Proving Ground

Proving Ground is a staking-plan backtester. It takes a run of already-decided bets — their odds and whether each won or lost — and replays them through 12 different staking plans at once, so you can see which plan would have made the most from the same bets, and which would have busted along the way.

It does not pick selections and it does not predict the future. It answers one question: given how you actually bet, how would each way of sizing your stakes have performed? Crucially, it ignores the stake you originally used — it only needs each bet’s odds and result, then re-stakes that same sequence under every plan.

• My Bet Tracker — replays the settled bets you’ve logged in the Bet Tracker (signed in). The most useful source, because it’s your real betting. (Pro)
• Import a file — upload a CSV or Excel export from a spreadsheet or bookmaker; you map the odds and result columns. Feeds the backtest only — it never adds anything to your Bet Tracker. (Pro)
• Synthetic — generate a realistic run of bets from a strike rate, average odds and spread. Great for trying the tool or stress-testing a plan at a chosen win rate. (Free)

You need a sequence of settled bets with decimal odds and a result (win / loss / void / half / push). Order matters — staking plans react to wins and losses as they happen — so keep the bets in the order you placed them. Set a starting bank and, if your exchange charges it, a commission, so the effective odds are right.

A backtest is most meaningful over a decent sample: a few hundred bets shows a plan’s behaviour far better than a dozen. The same bets are used for every plan, so the comparison is always like-for-like.

The plans fall into four families, from steady to volatile. You can run any mix of them; each has a sensible default so you can select and run with zero setup.

• Flat & proportional — Level stakes (a fixed % of your starting bank), Percentage of bank (a fixed % of your current bank), Compound bank (a twentieth of the bank), and fractional Kelly (sized from an assumed edge).
• Recovery — Resurgence and The Mint (the platform’s own recovery engines) and Stop-at-winner: these lift stakes through a losing run to claw it back.
• Progression — Regal (profit-spread), D’Alembert, Fibonacci and Labouchère: structured step-ups and step-downs.
• Martingale — doubles after every loss. It carries a permanent “high ruin risk” badge and is included as the cautionary tale, not a recommendation.

1. 1Data tab — choose a source (Synthetic, My Bet Tracker, or Import a file) and load the dataset. A summary card shows the bet count, realised strike rate and average odds.
2. 2Plans tab — tick the plans to compare. Open a plan’s drawer to adjust its one or two knobs, or leave the defaults.
3. 3Run the backtest — every selected plan is replayed over the same bets (off the main thread, so the page stays responsive; you can cancel a long run).
4. 4Results tab — read the ranked table; plans that busted sink to the bottom.
5. 5Compare tab — overlay the equity curves to see not just where each plan finished, but how bumpy the ride was.

• Final bank & net P&L — where each plan ended from the same starting bank.
• ROI / POT — profit relative to total staked, and relative to your starting bank.
• Max drawdown — the worst peak-to-trough dip along the way. A bigger final bank with a deep drawdown is a bumpier, riskier ride.
• Max stake — the largest single bet a plan demanded; recovery and progression plans can balloon this.
• Longest losing run — the worst cold streak in the data (the same for every plan).
• Busted — whether the plan ran out of bank (and at which bet). A busted plan is ranked last no matter how it looked beforehand.

A single backtest is one ordering of your bets — and luck of the draw in that order can flatter or punish a plan. The shuffle test re-runs the same bets in hundreds of different orders and reports the spread: the median final bank, the unlucky (10th-percentile) and lucky (90th-percentile) cases, and how often each plan busts.

It’s the difference between “this plan won once” and “this plan wins reliably.” A plan with a high median and a low bust rate is robust; one that wins big in the best order but busts often is just gambling on sequence.

Save a backtest as a named experiment (Pro) and it stores your data setup, the plans and the seed — not the numbers. When you reload it the results are recomputed from scratch, which is also a neat proof that the engine is deterministic: same inputs, same answer every time.

Export the results table to CSV or Excel (Pro) to keep or analyse elsewhere. The Excel file keeps money and percentages as real numbers, so you can sort and total them in the spreadsheet.

Free runs synthetic datasets through two plans (Level stakes and Percentage of bank) — enough to see how the tool works. Pro unlocks all 12 plans, your real Bet Tracker history and file import as data sources, the shuffle test, saved experiments, and CSV / Excel export.

Does Proving Ground take results from my other calculators?

Not as data. Its real-data source is the Bet Tracker (the bets you log there). It does include several tools’ staking methods — Resurgence, The Mint, Regal, Compound and Kelly — as plans you can compare, but it doesn’t read figures out of those calculators.

Does it use the stake I originally placed?

No. It deliberately ignores your original stake and re-sizes every bet under each plan; it only needs each bet’s odds and result.

Will the best backtested plan win for me next time?

No staking plan changes whether a bet wins or removes the house edge. The backtest shows how plans behaved on a past or synthetic run — use it to understand risk, not as a guarantee.

Why did a plan “bust” even though my bets made money overall?

Progression and recovery plans raise stakes through losing runs. A long enough cold streak can exhaust the bank before the wins arrive — which is exactly the risk the tool is built to expose.

A staking plan changes how much you stake, not whether a bet wins, and nothing removes the house edge. These tools are educational calculators — they do not place bets, take money, or give tips.

Set a daily loss limit and, if you need a break, a cool-off in Settings. Never stake money you cannot afford to lose. Free, confidential support: National Gambling Helpline 1800 858 858 (24/7).