Methodology
What a Backstedge result is made of.
Where the matches and odds come from, what a criterion is allowed to know, how a result should be read, and how the library strategies are chosen. Nothing here is a promise of future returns.
- Pre-match data only
No information from after kick-off ever enters the criteria.
- Identified odds
Real and estimated odds are always told apart.
- Reproducible computation
The same rules applied to the same data produce the same result.
- Limits flagged
Periods or markets with partial coverage are explicitly indicated.
1. Data coverage
Backtests are computed on real matches; when bookmaker odds are missing, the result says so instead of hiding it.
| Competition | Seasons | Matches |
|---|---|---|
| 2. Bundesliga | 2021/222026/27 | 1,629 |
| AFC Champions League Elite | 2021/222026/27 | 744 |
| Bundesliga | 2021/222026/27 | 1,621 |
| Championship | 2021/222026/27 | 2,940 |
| Community Shield | 2021/222026/27 | 6 |
| CONCACAF Champions League | 2021/222026/27 | 240 |
| CONMEBOL Libertadores | 2021/222026/27 | 927 |
| CONMEBOL Sudamericana | 2021/222026/27 | 941 |
| Copa del Rey | 2021/222026/27 | 702 |
| Copa Do Brasil | 2021/222026/27 | 764 |
| Coppa Italia | 2021/222026/27 | 253 |
| Coupe de France | 2021/222026/27 | 987 |
| Cup | 2021/222026/27 | 1,529 |
| DBU Pokalen | 2021/222026/27 | 640 |
| DFB Pokal | 2021/222026/27 | 363 |
| Eliteserien | 2021/222026/27 | 1,406 |
| Eredivisie | 2021/222026/27 | 1,707 |
| FA Cup | 2021/222026/27 | 4,737 |
| FA Cup | 2021/222026/27 | 373 |
| FIFA Club World Cup | 2021/222025/26 | 85 |
| Jupiler Pro League | 2021/222026/27 | 1,693 |
| KNVB Beker | 2021/222026/27 | 579 |
| La Liga | 2021/222026/27 | 2,019 |
| League Cup | 2021/222026/27 | 548 |
| Leagues Cup | 2021/222026/27 | 287 |
| Ligue 1 | 2021/222026/27 | 1,775 |
| Ligue 2 | 2021/222026/27 | 1,858 |
| MLS | 2021/222026/27 | 2,941 |
| NM Cupen | 2021/222027/28 | 823 |
| Premier League | 2021/222026/27 | 2,000 |
| Primeira Liga | 2021/222026/27 | 1,628 |
| Schweizer Cup | 2021/222026/27 | 363 |
| La Liga 2 | 2021/222026/27 | 2,448 |
| Serie A | 2021/222026/27 | 2,011 |
| Brasileirão | 2021/222026/27 | 2,239 |
| Serie B | 2021/222026/27 | 2,010 |
| Super Cup | 2021/222025/26 | 15 |
| Super Cup | 2021/222026/27 | 6 |
| Super Cup | 2021/222025/26 | 11 |
| Super Cup | 2021/222026/27 | 6 |
| Super Cup | 2021/222025/26 | 5 |
| Super Cup | 2021/222025/26 | 7 |
| Swiss Super League | 2021/222026/27 | 1,124 |
| Chinese Super League | 2021/222026/27 | 1,440 |
| Süper Lig | 2021/222026/27 | 1,848 |
| Superliga | 2021/222026/27 | 1,044 |
| Taça da Liga | 2021/222026/27 | 155 |
| Taça de Portugal | 2021/222026/27 | 867 |
| Trophée des Champions | 2021/222026/27 | 6 |
| Türkiye Kupası | 2021/222026/27 | 887 |
| Champions League | 2021/222026/27 | 1,368 |
| Conference League | 2021/222026/27 | 2,385 |
| Europa League | 2021/222026/27 | 1,217 |
| UEFA Super Cup | 2021/222026/27 | 6 |
| US Open Cup | 2022/232026/27 | 469 |
Coverage by market
Share of finished matches with real odds. Below that, the backtest uses estimated odds and flags it.
| Market | Real odds | Share of matches |
|---|---|---|
| 1X2 | since 2021/22 | 59% |
| Over / under goals | since 2021/22 | 59% |
| Both teams to score | since 2023/24 | 37% |
| Double chance | since 2023/24 | 35% |
| Half-time markets | since 2023/24 | 28% |
A backtest always shows the period actually covered by its bets, which can be shorter than the requested period if some seasons produced no bets.
2. How odds are chosen
A backtest uses closing odds; estimated odds are never presented as real odds.
- Closing odds
- The last odds available before kick-off. These are the odds used in backtests.
- Current odds
- The latest odds available for an upcoming match. They can still move.
- Estimated odds
- Used only when no historical bookmaker odds are available.
3. How criteria are computed
Every criterion uses only what was known before kick-off of the analyzed match.
- Form over 5 matches
- Requires the five most recent finished matches before the analyzed match — below that, the criterion cannot be computed.
- Wins or losses over the last 10 at the same venue
- Require ten full home or away matches; the window is never shrunk to whatever is available.
- Goal averages
- Computed over the ten most recent matches played before the analyzed one, all venues combined — ten required.
- Season win/loss %
- Percentages computed over every match of the current season, starting from three matches played.
- Standings
- Position known at the time of the match, in the current competition and season.
No data from the analyzed match or from any future match is used.
When the history is too short to compute a statistic, the match is discarded rather than guessed.
4. How live tracking works
Live tracking only records matches detected with real, up-to-date odds, then measures what the strategy actually delivers.
- 1Strategy activated
- 2Matching match detected
- 3Odds monitored
- 4Match played
- 5Result recorded
- 6Real performance updated
- To watch
- The match seems to qualify, but the final odds are not available yet: it will be re-evaluated as soon as new odds arrive. No alert and no live tracking until then.
- Detected
- The match currently matches the strategy, with real bookmaker odds available: live tracking will record this bet.
- Suspended
- The match qualified, but the available odds are too old or the odds feed is stale: it is no longer counted as detected until the feed is back.
A match postponed or canceled after detection is refunded: it counts neither in the profit nor in the win rate. See the upcoming matches of the library strategies.
5. How to read a result
An ROI only reads together with the number of bets, the worst drawdown and reliability; a past result guarantees nothing going forward.
- ROI
- Return relative to the total amount staked.
- Profit in units
- Profit with a flat 1 u stake per bet.
- Worst drawdown
- Largest drop between a peak and the following low.
- Reliability
- Overall assessment of how much confidence the result deserves.
- Stability
- Checks whether the result withstands small variations of the criteria.
More about reliability and stability
A backtest measures what a rule would have produced in the past, with a flat 1 u stake and no fees. It does not predict future results, and accounts neither for stake limits nor for the actual availability of the odds at the moment you would have bet.
The verdict, reliability and consistency rely on a single set of thresholds, identical on every page. Consistency is never called good on a single season, and a statistically promising result resting on too few bets is presented as needing confirmation. “Solid result” can only be reached once parameter stability has been tested and judged good.
Stability slightly shifts every numeric criterion, one at a time, and checks that the result stays consistent around the chosen values. It never hunts for “the best setting”: it tests the strength of the hypothesis.
6. How the library strategies are selected
The strategy library is a shortlist, not a prediction: the rules that stayed profitable on a season kept aside, out of many more that did not.
Each strategy comes from an exploration of many candidate rules over several seasons. A rule is kept only if its backtest stays profitable on a season that was never used to find it, with at least 200 bets and a stability analysis judged good.
The rules that lost, or that only worked on the seasons used to find them, are not listed. Reading only the survivors overstates how easy it is to find a profitable rule: that is why every strategy also shows its season-by-season results and its current season, including when it is losing, and why a current season with few bets is flagged as too small to judge.
The library is recomputed every day on the latest finished matches. A rule whose ROI turns negative leaves the list; its page stays readable at its address. With live tracking, Backstedge flags the upcoming matches that meet a strategy’s conditions; what to do with them is your call.