A change in the player’s history
Compare time, deposits and behaviour across snapshots. The original events remain inspectable.
XSLOT / LAB
Player laboratory
History, unit economics and the next scenario in one experiment.
Synthetic data · public context
Experiment configuration100 accounts · 365 days · mixed lifecyclesConfigure +
Saved runs
Completed runs appear here.
Runs are scoped to this browser session and retained for 24 hours. Download exports for durable copies.
Synthetic research workspace
Player lifecycle research
Explore the saved example or generate a cohort with your own settings.
Settings have changed. Results below belong to the previous saved run. Generate again to apply them.
Saved synthetic example
- Registered
- — Complete synthetic cohort
- First deposits · FTD
- — Depositors / registrations
- Deposits
- — Settled · complete history
- Cash GGR
- — After ledger corrections
- NGR · scenario
- — After gaming deductions
- CAC / registration
- — Configured acquisition cost
- LTV · scenario / registration
- — History + next 12 months
- D30 · observed
- — Active / mature on day 30
The whole player story
No cohort yet
Awaiting a history
SYNTHETIC HISTORYEvery signal begins with a story
Create a cohort to trace registrations, deposits, bonuses, gameplay and protection decisions.
Local computation · no paid APIsAccountMoneyPlayEngagementProtection
100%
Read along the wires. Solid line: recorded technical link; dashed line: event start order. Group periods may overlap. Spacing is not elapsed time. Animation illustrates event type, not additional activity.
Where observation ends, forecasting begins
Continue this history with a 7, 14 or 28 day forecast.
FARO / DECISION TRACE
From event to reasoned review
Inspect real model outputs on chronological snapshots, with the features that contributed to each signal.
FARO inference runs on demand. A technical signal is not a clinical diagnosis or a validated intervention outcome.
A signal with a traceable basis
Review model contributions alongside independent protection rules and missing data.
A documented operator decision
Demonstrate escalation for human review, promotion suppression or limit checks. Benefits and harm reduction require real-world evaluation.
Unit economics / explicit scenario
What is this cohort worth?
Saved history anchors the monthly margin. Change the assumptions to explore the remaining value.
From GGR to scenario value
Every amount below is per registration. Select a step to inspect the reason, then change the future assumptions.
Synthetic scenario · checking the saved calculation
IncomeDeductionsTotalsFuture assumptionEUR / registration
Linear monetary scale. X: calculation order. In 3D, depth is decoration; height remains money. Missing values stay unavailable.
History before CAC —+Future PV —=Scenario LTV —−CAC —=After CAC —
Exact calculation and all deductions
| Calculation layer | Amount / registration | Basis |
|---|
A reconciled scenario demonstrates the mechanism, not measured market LTV or ROI. Corporation tax, fixed licence fees, levies and overhead require separate operator settings; they are not silently assumed to be zero.
Value after acquisition cost
Observed history + next 12 months
—Per registered account
Loading the saved scenario…
Expected future activity
Share of registrations · scenarioAt least one activity in a future 30-day window; permitted accounts only. Counts are expected values, not exact future players.
Additional months to recover CAC—
LTV / CAC—
Remaining value before CAC—
How LTV and retention are calculated
Observed history
GGR = cash bets minus cash winnings, after corrections. NGR = GGR minus realised bonuses, configured gaming tax, provider/platform and affiliate costs. Acquisition cost is separate. ARPU below uses accounts that were active at least once.
Remaining scenario
Estimated LTV = observed contribution before CAC per registration + discounted future contribution. Future activity starts from the latest complete 30-day window and declines by the chosen monthly continuation rate. CAC is deducted once. A negative result remains negative.
Two retention definitions
D30 is exact-day observed activity among accounts that completed that day. Monthly continuation is activity in two adjacent complete 30-day windows divided by eligible active predecessor windows. These rates answer different questions.
Scope of the estimate
Synthetic histories and configured costs support scenario exploration. The model assumes the pooled mature cohort represents the future, constant margin and no explicit reactivation. The three sensitivity cases are assumptions, not confidence intervals.
Observed cohort · scenario economics
From first deposit to the economics of play
Loading a saved synthetic example. Generate a cohort to inspect your own experiment.
01
Observed milestones
Unique accounts across the complete observation window. Each stage is a subset of the previous one.
First deposit = at least one settled deposit. Repeat deposit = at least two. The denominator is registered accounts, not site visitors.
02
Who was active at each age?
Exact-day retention since registration. Each rate uses only accounts old enough to have completed that UTC day.
D0 is the UTC registration date. Activity means a settled deposit, a cash bet or a restricted-bonus bet. An inactive day does not establish churn.
active / mature accounts · a dash means unavailable, not zero
03
Where the contribution comes from
Cash GGR, realised cash bonus cost and configured operator expenses, in the cohort currency.
Contribution is a scenario calculation, not net profit or LTV. Payroll, hosting and unconfigured expenses are excluded. Unconverted restricted bonus grants are excluded from cash cost; their conversion into cash is included.
04
Which paths make up this cohort?
Assigned lifecycle scenarios, counted from account profiles. These are generation inputs, not discovered behavioural segments or risk diagnoses.
These views describe generated histories. They do not prove that a campaign caused retention, identify clinical risk or establish performance on real players. Every exported value retains its denominator and definition.
Daily totals and technical overview
Generated accounts—No run yet
Events & transactions—A traceable event ledger
FARO feature coverage— / 180Checked from the live catalog
Integrity checks—Measured after generation
Cohort over time
Daily totals reconstructed from the generated history.
The shape of your experiment
Generate a cohort to explore activity, cash flows and changes over time.
Operator view
Beyond FAROPayments, bonus economics, product mix and account restrictions will be calculated from events.
A history you can follow
Select an account to inspect its profile, daily activity and transaction lineage.
Event ledger
No account selected.
| Time · UTC | Event | Product | Amount | Balance after | Details |
|---|
Every feature, explained
FARO contract plus operational extensions. Inspect a definition to see its formula, required events, eligibility and value for the selected account.
What happened to the 27 unavailable features?
Previous gaps, what can be simulated explicitly, and what still requires operator evidence or a validated model. Availability does not establish behavioural realism.
Definitions are available before generation. Values appear after selecting an account.
The live generator catalog will appear here.
Forecast the same player history
The account selected in the journey continues here. The simulator creates history; models forecast its next 7–28 days. Return to the journey to see the future boundary.
Forecast settings changed. The visible comparison belongs to the previous settings.
Future daily values
No aligned model forecasts are available for this request. Inspect the model status below.
Does blending improve this scenario?
Lower MAE and RMSE mean smaller holdout errors. WAPE is relative absolute error and is undefined when actual totals are zero. Test targets never fit the weights.
| Model / blend | Current forecast | Horizon total | MAE | RMSE | WAPE | Test points |
|---|
Weights fitted in the past, then frozen
Chronological evaluation
Inspect protocol and calibration
Daily forecast values and model disagreement
The spread across model means describes model disagreement. It is not a calibrated prediction interval.
What this experiment measures
A comparison on synthetic history tests this scenario and this forecasting protocol. It does not establish performance on real players, responsible gaming validity, or the causal effect of an offer or traffic source. No pooled prediction interval is produced.
EVIDENCE → EXPLANATION
Local summary availableWhat happened. What supports it.
Read the same history through its events, financial records and saved model results. Every observation points back to evidence.
The local summary uses fixed rules. AI interpretation adds a narrative when the model is connected. Neither changes the source data.
Observed in the data
Questions for operator review
Evidence behind the explanation
Version and reproducibility
Interpretation describes already calculated evidence. Model feature contributions, forecast errors and synthetic assumptions keep their own meaning; fluent text is not additional validation.
Change one assumption. Inspect the difference.
Replay the same population, seed and observation window. Compare operational outcomes before designing a real pilot.
Baseline cohort
The baseline stays fixed while a variant is generated.
| Measure · whole cohort | Baseline | Variant | Difference |
|---|
A measured difference inside the simulator
A shared seed and matched inputs reduce configuration differences. Random event streams can still branch. This is not a causal effect, a real-player uplift or a revenue promise. A bonus, retention or VIP policy must respect protection rules before any commercial action.
A contract for your data team
Selected synthetic example, event fields, mapping template, feature definitions and pilot validation criteria.
Generated locally from this experiment. No operator data is uploaded or sent.
Can this run support the research?
Integrity is measurable. Behavioural realism still needs validation against real histories.
Generate a cohort to run integrity and eligibility checks.
What public data can establish
Published aggregates can constrain market totals and product context. They cannot establish individual behaviour, risk labels or causal offer effects. Scenario labels are simulation inputs, not diagnoses or trained ground truth.
EVIDENCE / PUBLIC DATA
Know what shapes the simulation
Trace each assumption to its source, population and period. A source count is not a measure of behavioural realism.
Waiting for the source catalogue
Explore a market
The catalogue will show the evidence available for each market.
Why Spain has a separate calibration · what the grades mean
Partial joint distribution
DGOJ publishes annual deposit frequency × mean deposit amount. The existing ES sampler uses these bins only with an explicit share of activity at one operator. Timing and transitions remain assumptions.
Aggregate comparison targets
GB, NL and DK offer selected session, loss or return metrics. They can support comparisons, but are not fitted player trajectories and do not currently change the sampler.
Context and definitions
Market reports and regulatory definitions frame scenarios. Revenue, stakes, accounts and people have different denominators; market totals cannot become player probabilities.
These grades describe the available evidence, not model quality. Public aggregates do not establish individual risk labels, VIP affordability or the causal effect of retention actions.
SOURCE LIBRARY
Publications & extracted constraints
RECOMPILE / PREVIEW
Bring new evidence into the research
Download the registry as a template, edit its sources and numeric constraints, then upload the JSON. Compilation checks structure and provenance without fetching the linked websites.
The preview is separate from active generation. New records require source review and a reviewed deployment before they can influence the simulator.
Your file will be sent only when you compile it.
Validation details
Saved run assumptions and complete source list
Evidence and assumptions
Keep the original population, period and measurement unit attached to every public source.
Public reference material
Loading the source registry.
A report you can inspect and share
Each export retains synthetic labels, configuration and provenance.
Xslot Research
Synthetic cohort research report
Generate an experiment to build its report.
Run manifest & model capabilities
No completed run.
Turn this experiment into a pilot
Define one task for your platform, the data it needs and how you will accept the result. Take a brief into your next product or engineering meeting.