Civil Law and UAE: Probabilistic Justice and Threshold-Based Adjudication

1. Introduction

Probabilistic justice refers to judicial decision-making where the available evidence does not provide absolute certainty, so the court determines disputed facts by evaluating the relative probability of competing explanations.

Threshold-based adjudication refers to situations where the law establishes a particular evidentiary or procedural threshold that must be satisfied before a court can grant a remedy, impose liability, permit an application, or make a particular finding.

These concepts are particularly relevant to modern UAE civil litigation because disputes increasingly involve:

incomplete digital evidence;

complex financial transactions;

multiple possible causes of loss;

algorithmic evidence;

expert evidence;

fraud allegations;

insurance claims;

probabilistic financial calculations;

urgent interim applications; and

technologically complex disputes.

A fundamental distinction should be maintained:

Probabilistic adjudication does not mean that UAE courts decide cases by guessing probabilities. It means that legally relevant facts are established by applying the applicable standard of proof to the available evidence.

The most important civil standard in the DIFC authorities is the balance of probabilities—whether a proposition is more likely than not. (DIFC Courts)

The current onshore UAE general civil-law framework is Federal Decree-Law No. 25 of 2025 on the Civil Transactions Law, which came into force on 1 June 2026 and replaced the former 1985 Civil Transactions Law.

2. Meaning of probabilistic justice

The expression can be broken down into three ideas.

A. Evidence is often incomplete

Civil courts rarely have perfect information.

For example:

A company claims that a manager diverted AED 10 million.

There may be:

bank records;

emails;

WhatsApp messages;

accounting discrepancies;

witness testimony;

incomplete documents; and

competing explanations.

The court must decide what probably happened.

B. Competing explanations are compared

The court may ask:

Is the claimant's explanation more probable than the defendant's explanation?

C. A legal threshold must be satisfied

The court does not award judgment simply because one explanation is conceivable.

The applicable legal threshold must be satisfied.

Thus:

Evidence → Probability assessment → Legal threshold → Finding → Legal consequence

3. Probabilistic justice is not statistical justice

This distinction is extremely important.

Statistical probability

A mathematical proposition such as:

There is a 75% probability that Event A caused the damage.

Judicial probability

A legal conclusion such as:

On the evidence, Event A is more likely than not to have occurred.

The second does not necessarily require mathematical percentages.

A judge may consider:

witness credibility;

documents;

chronology;

expert evidence;

commercial context;

circumstantial evidence;

alternative explanations; and

inherent probabilities.

Therefore:

Judicial probability is evidential reasoning, not merely mathematical calculation.

4. Threshold-based adjudication

A threshold is the level that must be reached before a legal consequence follows.

Different procedural situations may have different thresholds.

For example:

The important point is:

Not every stage of litigation uses the same decision-making question.

5. Balance of probabilities

The ordinary civil standard is generally expressed as:

More likely than not.

In Graciela Ltd v Giacobbe [2014] DIFC CFI 027, the DIFC Court expressly stated that the burden of proof was on the claimant and that the civil standard was the balance of probabilities. (DIFC Courts)

The Court explained that the question is whether, on the evidence, the court is satisfied that the event occurred.

Thus:

51% likelihood is conceptually enough to satisfy a binary more-likely-than-not test, but courts do not ordinarily need to assign numerical percentages.

6. Serious allegations do not create a separate civil standard

A particularly important principle is that a serious allegation does not automatically require a legally higher standard of proof.

GFH Capital Ltd v David Lawrence Haigh [2014] DIFC CFI 020

The case involved serious allegations of dishonesty.

The DIFC Court confirmed that the standard remains the balance of probabilities. The seriousness or inherent improbability of an allegation can affect how the evidence is evaluated, but it does not create a separate "heightened" civil standard. (DIFC Courts)

Principle

The court effectively asks:

What does the totality of the evidence establish?

It does not formally change:

"balance of probabilities"

into:

"beyond reasonable doubt."

This is fundamental to probabilistic justice.

7. Case Law 1 — Graciela Ltd v Giacobbe [2014] DIFC CFI 027

This case is a leading illustration of civil probabilistic reasoning.

The Court expressly explained the civil standard as the balance of probabilities.

It also discussed the relationship between seriousness and probability.

Principle

The more inherently improbable a factual allegation appears, the stronger the evidence may need to be before the court is persuaded that it occurred.

But:

The legal standard remains the same.

Significance

This establishes the basic architecture:

Evidence → evaluation of inherent probabilities → balance of probabilities → factual finding.

(DIFC Courts)

8. Case Law 2 — SBM Bank (Mauritius) Ltd v Renish Petrochem FZE & Mehta [2022] DIFC CA 011

This Court of Appeal decision is particularly valuable.

The Court stated that the standard of proof in civil proceedings remains the balance of probabilities.

It also explained that allegations of dishonesty can be established through inference from primary facts.

The relevant question is whether the inference of dishonesty is more probable than an innocent explanation. (DIFC Courts)

Why this matters

Fraud is rarely established by a document saying:

"I intend to commit fraud."

Instead, the court may consider:

unusual payments;

inconsistent explanations;

suspicious timing;

financial records;

communications;

conduct before and after the event.

The evidence is assembled into a probabilistic picture.

Important principle

Circumstantial evidence can satisfy a civil evidentiary threshold when the totality of the evidence makes the relevant inference more probable than the competing explanation.

9. Case Law 3 — GFH Capital Ltd v David Lawrence Haigh [2014] DIFC CFI 020

This case deserves separate treatment because it demonstrates the relationship between serious allegations and evidentiary thresholds.

The defendant faced allegations involving dishonesty and fraudulent conduct.

The Court rejected the idea that a separate higher standard of proof applies merely because the allegation is serious.

Instead, the court considers the inherent probability of the allegation when assessing the evidence. (DIFC Courts)

Example

Suppose:

Explanation A: ordinary accounting mistake.

Explanation B: deliberate diversion of funds.

If the evidence is equally consistent with both, the serious allegation may not be established.

But if:

payments went to related entities;

records were deliberately altered;

instructions were concealed;

explanations were contradictory; and

there was no credible innocent explanation,

the cumulative evidence may cross the civil threshold.

10. Case Law 4 — Amira C Foods International DMCC v IDBI Bank Ltd [2018] DIFC CFI 027

This is one of the most useful authorities for probabilistic causation.

The Court considered how causation may sometimes be inferred where direct evidence of the precise causal mechanism is unavailable.

The Court emphasised that alternative theories of causation must be considered. If the defendant's alternatives are improbable, that may strengthen the inference that the proven negligence caused the loss. (DIFC Courts)

Principle

The court may reason:

Defendant's negligence is proved.

Damage occurred.

The damage is consistent with that negligence.

Alternative explanations are considered.

Alternatives are less probable.

The defendant's conduct becomes the more probable explanation.

This is an excellent example of probabilistic civil reasoning without statistical certainty.

11. Case Law 5 — Haya Spa LLC v Harper Real Estate / Hasan Real Estate [2016] DIFC SCT 150

The case involved a negligence claim and the DIFC Court explained the elements of civil liability.

The Court's analysis included:

duty;

breach;

causation; and

damage.

It referred to the requirement of establishing the relevant causal connection, including the "but for" and substantial-cause concepts under the DIFC framework.

The case also demonstrates that intervening events, claimant conduct and mitigation can affect the ultimate result.

Importance

Threshold-based adjudication does not stop at:

"The defendant probably did something wrong."

The claimant must cross each necessary legal threshold:

Duty → Breach → Causation → Damage

Failure at one stage can defeat the claim.

12. Case Law 6 — Ludiala v Lucaan Ltd [2020] DIFC SCT 139

This case demonstrates the failure to cross the causation threshold.

The claimant alleged psychological injury connected with workplace circumstances.

The evidence did not sufficiently establish the necessary causal relationship between the defendant's conduct and the claimed psychiatric condition.

Principle

A claimant cannot simply establish:

Defendant's conduct occurred.

and then assume:

Therefore, defendant caused my injury.

The causal bridge must be proved.

Significance

This demonstrates the negative side of probabilistic justice:

Where competing causal explanations remain unresolved and the claimant cannot establish the necessary causal connection on the applicable standard, the claim may fail.

13. Case Law 7 — BAM Higgs & Hill LLC v Affan Innovative Structures LLC & Amer Affan [2021] DIFC CFI 106

This construction dispute illustrates claim-by-claim threshold analysis.

The Court considered numerous alleged breaches and whether particular breaches actually caused the losses claimed.

The Court referred to the requirement of a causal connection between fault and damage and noted that not every alleged breach necessarily caused the claimed loss. (DIFC Courts)

Importance

Suppose a claimant alleges:

10 contractual breaches;

AED 20 million of damage.

The court does not simply say:

"There were breaches, therefore AED 20 million is recoverable."

Instead:

Breach 1 → Loss 1?
Breach 2 → Loss 2?
Breach 3 → Loss 3?

This is threshold-based adjudication at the level of individual heads of loss.

14. Case Law 8 — IDBI Bank Ltd v Amira C Foods International DMCC [2019] DIFC CA 014

The Court of Appeal examined causation in the context of damages and benefits.

A benefit should be taken into account against damages only where it is causally connected with the breach or arises from mitigation of its consequences.

Principle

Causation operates not only on the claimant's loss but also on claimed benefits.

Thus:

Loss must be causally connected to the wrong before recovery.

And:

A benefit must be causally connected before it can appropriately reduce the damages.

This demonstrates the importance of maintaining a precise causal chain rather than using broad assumptions.

15. Case Law 9 — Lals Holding Ltd v Emirates Insurance Company [2022] DIFC CFI 073

This insurance dispute involved complex questions concerning pandemic-related business interruption and insurance coverage.

The Court stated that the burden of proof, to the required standard, rested on the claimants and that the applicable standard was the balance of probabilities. It also observed that what is necessary to satisfy that burden is fact-specific. (DIFC Courts)

Importance

This is particularly useful for threshold-based adjudication.

There is no universal evidentiary checklist that guarantees success.

The threshold must be evaluated against:

the particular legal issue;

the relevant contractual wording;

the evidence;

the factual circumstances.

16. Case Law 10 — Arqaam Capital Ltd v DFSA [2012] DIFC CFI 006

This case demonstrates the distinction between a preliminary regulatory assessment and a final adjudicative finding.

The judgment discussed a regulatory standard involving whether it appeared reasonably likely that a breach had occurred, contrasted with the Financial Markets Tribunal's determination of whether the alleged breach was proven on the balance of probabilities. (DIFC Courts)

Importance

This provides an excellent example of different thresholds for different procedural stages.

A preliminary regulator may need enough information to justify further action.

A final adjudicator may need proof sufficient to establish the breach under the applicable civil standard.

Therefore:

A preliminary probability threshold is not necessarily equivalent to the final merits threshold.

17. Final adjudication versus interim adjudication

This distinction is central.

Final judgment

The court determines:

Has the claimant proved the necessary facts on the applicable civil standard?

Interim stage

The court may instead ask:

Has the applicant satisfied the procedural requirements for temporary relief?

The purpose is different.

An interim injunction does not necessarily establish that the applicant will ultimately win the case.

Similarly:

A freezing order does not constitute a final finding of liability.

This is an important safeguard against confusing procedural thresholds with substantive adjudication.

18. Thresholds can be issue-specific

A single lawsuit may contain many different thresholds.

For example:

Jurisdiction threshold

Does the court have jurisdiction?

↓

Standing threshold

Is the claimant entitled to bring the claim?

↓

Evidentiary threshold

Has the relevant fact been proved?

↓

Liability threshold

Have the elements of the cause of action been established?

↓

Causation threshold

Did the defendant's conduct cause the loss?

↓

Damages threshold

Has the amount of loss been proved?

↓

Remedy threshold

Is the requested remedy legally available?

This creates a layered adjudication model.

19. Threshold-based adjudication and evidence

The quality of evidence becomes crucial.

Strong evidence

contemporaneous documents;

authenticated electronic records;

independent expert evidence;

reliable financial records;

consistent witness testimony;

objective transaction records.

Weaker evidence

speculation;

unsupported assumptions;

hearsay without sufficient foundation;

retrospective reconstruction;

unexplained estimates;

inconsistent testimony.

The court's probabilistic assessment is therefore only as good as the evidence on which it is based.

20. Circumstantial evidence

Probabilistic justice does not require direct evidence in every case.

Consider:

A → transaction
B → hidden ownership
C → unusual payment
D → altered records
E → contradictory explanation

No individual fact may prove fraud.

But together they may support a sufficiently strong inference.

This is why SBM Bank is important: the court recognised that fraud or dishonesty is often established through inference from primary facts rather than direct evidence. (DIFC Courts)

21. Cumulative evidence

One of the most important principles is:

Evidence should often be considered cumulatively rather than atomistically.

For example:

Individually, each may be insufficient.

Together, they may cross the applicable threshold.

This is particularly important in:

fraud;

fiduciary breaches;

corporate disputes;

money diversion;

cybersecurity;

digital transactions.

22. Probabilistic justice and expert evidence

Modern civil disputes increasingly depend on experts.

Medical

Probability of injury from particular treatment.

Construction

Probability that a defect caused structural damage.

Finance

Probability that a particular event caused financial loss.

Insurance

Probability that a particular event falls within an insured peril.

Technology

Probability that a system failure caused data loss.

Digital assets

Tracing probability through multiple wallets and transactions.

The expert's function is to provide technical evidence.

The court's function remains:

Determine the legal consequence of the evidence.

23. AI and probabilistic adjudication

The emergence of AI makes this topic particularly important.

AI systems can estimate:

litigation risk;

likely damages;

similarity between cases;

document relevance;

probability of particular outcomes.

But predictive analytics is not judicial adjudication.

A statistical model might say:

Similar cases produced damages between AED 100,000 and AED 150,000.

A judge must still determine:

whether the legal elements are established;

which evidence is admissible/reliable;

whether the precedent is applicable;

whether the facts are sufficiently similar;

whether the claimant proved actual damage.

Therefore:

Prediction can assist legal analysis, but prediction is not itself a judicial finding.

24. Algorithmic thresholds

Platform disputes provide another example.

An algorithm may say:

Risk score > 90 → suspend account.

This is a technical threshold.

But the legal threshold may be different.

The court may ask:

Was suspension contractually authorised?

Was the underlying data accurate?

Was the decision made according to the contractual terms?

Did the platform breach another legal obligation?

Thus:

Algorithmic threshold ≠ legal threshold

This distinction will become increasingly important in AI-related civil litigation.

25. Probabilistic causation

Probabilistic justice is particularly relevant to causation.

Suppose:

A = defective product
B = improper maintenance
C = extreme weather
D = claimant's conduct

All four potentially contributed to damage.

The court must determine:

Did A contribute?

Did B contribute?

Did C constitute an intervening event?

Did D contribute?

What loss is attributable to A?

What reduction or limitation follows?

The decision is not necessarily mathematical.

It is a legal assessment of evidence and causal responsibility.

26. Thresholds and burden of proof

The burden of proof and standard of proof should be distinguished.

Burden of proof

Who must establish the proposition?

Standard of proof

How convincing must the evidence be?

For example:

Claimant bears burden → balance of probabilities → factual proposition established.

In Lals Holding, the Court expressly placed the burden on the claimants and identified the balance of probabilities as the applicable civil standard. (DIFC Courts)

27. Burden may differ by issue

A single case may involve different burdens.

For example:

Claimant

Must prove:

contract;

breach;

damage;

causation.

Defendant

May have to establish:

a contractual defence;

payment;

limitation;

contributory conduct;

another affirmative defence.

Therefore, threshold-based adjudication requires careful identification of which party must establish which proposition.

28. Evidential thresholds and preliminary applications

Modern litigation contains many applications that do not ask:

"Who ultimately wins?"

Instead, they ask:

Should documents be disclosed?

Should assets be frozen?

Should a witness be cross-examined?

Should evidence be admitted?

Should a case proceed to trial?

Should judgment be entered summarily?

For example, in Jonathan Lau v Qashio Holding Company Ltd [2026] DIFC CFI 058, the Court explained that an application for pre-action production involves first determining whether the procedural conditions are satisfied and then deciding whether the court should exercise its discretion. The requirement that parties be "likely" to be involved in later proceedings was not equated with proving that proceedings were more probable than not. (DIFC Courts)

Importance

This demonstrates:

Different legal questions can legitimately use different thresholds.

29. Threshold-based adjudication and access to justice

Thresholds serve important functions.

They prevent:

purely speculative claims;

premature remedies;

unsupported allegations;

unnecessary interference with property;

excessive litigation costs.

But thresholds must also be applied fairly.

An excessively demanding threshold could make legitimate claims impossible to prove.

Therefore, the objective is:

Sufficient proof without demanding impossible certainty.

30. Thresholds and digital evidence

Digital disputes create another challenge.

Suppose a claimant produces:

screenshot;

WhatsApp conversation;

blockchain transaction;

email;

server log.

The court may need to determine:

authenticity;

attribution;

integrity;

relevance;

reliability;

probative value.

The fact that a digital record exists does not automatically establish the truth of every proposition contained within it.

Thus:

Digital evidence must still cross the relevant evidentiary threshold.

31. Fraud and probabilistic justice

Fraud provides one of the clearest examples.

Fraud is rarely proved through direct evidence.

Instead:

Primary facts → inference → competing explanation → probability assessment → finding

SBM Bank expressly recognised this inferential structure. (DIFC Courts)

The court can consider the evidence as a whole rather than asking whether every individual fact independently proves dishonesty.

32. The "no percentage required" principle

Courts generally do not need to say:

"The claimant has a 72% probability of being correct."

The legal test is ordinarily qualitative:

more likely than not.

Consequently, probabilistic justice should not be confused with a numerical scoring system.

The judge is not necessarily calculating:

Probability = 0.67

Instead, the judge evaluates the evidentiary record and reaches a legally justified finding.

33. Probabilistic justice and judicial discretion

Probability does not mean unlimited judicial discretion.

A court must ordinarily:

identify the relevant legal rule;

identify the burden of proof;

evaluate admissible/reliable evidence;

give reasons;

consider material competing explanations;

apply the law to the established facts.

The judgment therefore remains legally structured.

34. Current UAE legal-system perspective

The UAE is particularly interesting because it has multiple civil-law environments.

Mainland UAE

Current general civil law:

Federal Decree-Law No. 25 of 2025

DIFC

Separate common-law-influenced statutory framework and procedural rules.

ADGM

Separate legal framework with its own courts and rules.

Consequently, a statement about "UAE probabilistic justice" should not automatically assume that every evidentiary rule is identical across:

mainland UAE;

DIFC; and

ADGM.

The DIFC authorities cited above are therefore best understood as illustrative UAE-region authorities, unless the particular dispute is governed by DIFC law.

35. Important cases at a glance

36. A practical threshold model

A useful model for UAE civil litigation is:

Stage 1 — Legal threshold

Is there a recognised legal claim?

↓

Stage 2 — Evidential threshold

Has the necessary factual proposition been proved?

↓

Stage 3 — Causal threshold

Did the defendant's conduct cause the relevant loss?

↓

Stage 4 — Damage threshold

Has actual legally recoverable damage been established?

↓

Stage 5 — Remedy threshold

Is the requested remedy legally available?

This can be represented as:

Legal rule → Evidence → Probability → Threshold → Finding → Remedy

37. Safeguards against arbitrary probabilistic justice

A legitimate probabilistic system requires safeguards.

1. Reasoned judgment

The court should explain why one explanation was accepted over another.

2. Evidence-based reasoning

Probability should arise from evidence, not intuition alone.

3. Alternative explanations

Material alternative explanations should be considered.

4. Correct legal standard

The court should not silently replace the applicable standard with another.

5. Burden identification

The court should identify which party bears the burden.

6. Separation of interim and final findings

A preliminary threshold should not automatically become a final merits finding.

7. Expert scrutiny

Technical evidence should be evaluated rather than blindly accepted.

38. Probabilistic justice in construction disputes

Consider:

Contractor delay + subcontractor delay + government approval delay + extreme weather.

The claimant demands AED 20 million.

The court may ask:

Which delay was attributable to the defendant?

Which delays overlapped?

Which delay was on the critical path?

Would the project have been delayed anyway?

What loss was actually caused?

What part was mitigated?

This is not merely a question of calculating probability.

It is:

probability + causation + contractual allocation + evidence + legal threshold.

39. Probabilistic justice in financial disputes

Consider:

Misrepresentation → investment → market collapse → partial recovery.

The claimant says:

"The defendant caused all my losses."

The defendant says:

"The market caused the loss."

The court may construct a counterfactual:

What would the claimant probably have done if the misrepresentation had not occurred?

This type of counterfactual reasoning is reflected in DIFC financial litigation such as Alawwal Capital JSC v Rasmala Investment Bank Ltd [2025] DIFC CFI 038.

40. Probabilistic justice in AI disputes

Consider:

Faulty algorithm → incorrect risk score → account suspension → loss of customers → reduced revenue.

There may be uncertainty at every stage.

The court might need to determine:

Was the algorithm defective?

↓

Was the platform responsible?

↓

Was suspension authorised?

↓

Did suspension cause customer loss?

↓

Would customers have left anyway?

↓

What amount of loss is attributable to the suspension?

This is an archetypal modern probabilistic adjudication problem.

41. Main legal principles

Principle 1

Civil adjudication generally works through evidentiary probability rather than absolute certainty.

Principle 2

Balance of probabilities remains the ordinary civil standard in the cited DIFC authorities.

Principle 3

Serious allegations do not automatically create a higher legal standard.

Principle 4

The inherent probability of an allegation can affect how evidence is evaluated.

Principle 5

Circumstantial evidence can establish facts through inference.

Principle 6

Causation can sometimes be inferred after competing explanations are examined.

Principle 7

Different procedural stages can employ different thresholds.

Principle 8

A preliminary threshold should not be confused with final liability.

Principle 9

Each claimed loss must satisfy its own causal and evidentiary requirements.

Principle 10

Probabilistic reasoning does not permit speculation.

42. Conclusion

Probabilistic justice and threshold-based adjudication in UAE civil law describe an evidence-driven approach in which courts make legally structured decisions despite incomplete information.

The court does not demand absolute certainty in ordinary civil litigation. Instead, it determines whether the relevant proposition has been established under the applicable civil standard. The DIFC authorities repeatedly identify the balance of probabilities as the ordinary civil standard. (DIFC Courts)

At the same time, different stages of litigation can involve different thresholds. Arqaam Capital illustrates the distinction between a preliminary regulatory assessment and final proof of a breach, while Jonathan Lau v Qashio demonstrates that a procedural requirement using the word "likely" is not automatically a balance-of-probabilities test. (DIFC Courts)

The central safeguards are therefore:

Evidence + correct burden + correct standard + consideration of alternatives + reasoned findings + correct legal threshold.

The concept becomes particularly important for AI, digital evidence, complex financial transactions, insurance, construction, fraud, cybersecurity and multi-factor causation, where courts increasingly have to make legally justified findings from incomplete and technically complex evidence.

One-Minute Revision

Probabilistic Justice

Decision-making under incomplete evidence.

Not mathematical guessing.

Usually involves comparative evaluation of competing explanations.

Balance of probabilities = more likely than not.

Threshold-Based Adjudication

A legal consequence follows only after a prescribed threshold is satisfied.

Different procedural stages may have different thresholds.

Interim threshold ≠ final liability.

Key cases

Graciela → balance of probabilities.

GFH Capital → serious allegation does not create a separate higher standard.

SBM Bank → fraud can be inferred from cumulative primary facts.

Amira C Foods → causation can be inferred after considering alternative explanations.

Haya Spa → duty + breach + causation + damage.

Ludiala → insufficient causation defeats claim.

BAM Higgs → each loss needs causal connection.

IDBI v Amira → causal connection matters to damages/benefits.

Lals Holding → burden and standard are fact-specific.

Arqaam Capital → preliminary and final thresholds can differ.

Core formula:

Evidence → Probability Assessment → Applicable Threshold → Factual Finding → Legal Consequence → Remedy