We have three main sections sections in the Backtest page: the filter settings, the summary and the results. Let's take a look at them.
Backtest Filters
When loading the Backtest page, the Filters section shows your current filter settings. You can see the filter settings for each of the tournament levels when you click on the given tournament level at the top.
β
We currently predict the Match Winner market. There are 6 Levels (ATP, ATP Challenger, ITF Men, WTA, ITF Women 25k+ and ITF 15k), which you can select either independently or together (using All Markets/All Levels).
You can enabled/disable Each Marker-Level combinations with the toggle button and you can change the filter sets for each of these combinations.
A filter set is a group of settings (Odds range, Probability range and Expected Value range) tied to every market and tournament level. The tips that are visible in the app or the predictions pages are controlled by your filter set: only the tips matching the filter set are shown.Β
Note: when first signing in to our app or website, you automatically get the default filter set assigned. You can change the filter set later on, on the Backtest page.
The range sliders can be used to limit what kind of tips are shown on the backtest and ultimately as future tips:
Odds Range: it controls the odds range where a tip is taken into account. If a prediction's odds are out of this range, they are not considered when backtesting and not shown as prediction.
Probability Range: it controls the probability range where a tip is taken into account. If a tips's probability is not matching this filter, it won't be taken into account.
Expected Value (EV) Range: it controls if a tip with a given EV should be taken into account or not. (Note: for an explanation of EV please see the article How to interpret predictions.) Important to note that if the EV slider is set to -50% to +50%, it practically disabled EV-based filtering, thus tips with any EV, even tips without EV are taken into account
The next part shows common settings:
Model version: shows the current model version.
Starting Bankroll: the bankroll we start backtesting with.
Flat Stakes/Percentage Based Stakes selector: it is used to set flat stakes or stakes proportional to the total bankroll.
Flat Stakes: you can specify the size of the stake
Percentage Based stakes: you can specify the Percent of Bankroll Risked and the Max Stake
Date Range: start and end dates we want the backtest to run
The buttons:
Apply Changes: applies all changes and creates the summary table, charts and the results table.
Revert to Recommended Settings: reverts to the recommended settings. Make sure to click Save afterwards.
Save All: saves your filter settings to your account. When saved, all settings will be saved to your account and future tips will be presented to you based on these settings. These settings will apply to the tips on the website, mobile apps and email notifications as well.
Backtest Results Data
This is a summary of the backtest results showing the most important metrics. The results are affected by the Filter and Common settings.
Yield: the yield produced by the tips. Yield = Total profit / Total stakes amount.
Profit: the profit the settings have generated over the period.
ROI: the Return of Investment. ROI = Total profit / Staring bankroll
Success Rate: the success rate (or hit rate) for the tips. Success Rate = Number of tips won / Total number of tips
Max Drawdown: the maximum drawdown during the period. It is the biggest drop of the bankroll at any given time compared to previous heights and is expressed by the percentage of the previous height.
Number of Tips: the total number of tips
P-Value: It's a term used in statistics. P-Value shows the probability of the shown data being the result of pure luck. The lower the P-Value the better. The range of P-Value is 0% to 100%.
Backtest Results Chart
The chart shows how the Total Profit and profits for the individual levels looked during the period set. You can hover over the chart to see values at selected dates.
Backtest Table of Results
This last section shows all the tips provided by the model.
You can search using the "Search" field to the right
You can also see two buttons to export the data to the clipboard to to CSV. Exporting usually takes a lot of time, especially if the selected ranges are wider than defaults.
See Backtesting Pitfalls for the mistakes you can make while backtesting and how to avoid them.