We are not a tipster service. We are a research firm that happens to publish its work.
Our goal is to give you the same information a professional trader would have — clearly presented, without hype, and without telling you what to do.
Tennis
Football
How the Model Works
L1 · L2 · Discretionary

Layer 1 (L1) — A serve and return model that estimates each player's true win probability based on how they perform on their own serve and on the return, calibrated to the specific surface.


Layer 2 (L2) — An Elo model that tracks player ratings over time, updating after every match. It captures long-run skill levels and surface-specific strengths that shorter-term data alone can miss.


Ensemble & Discretionary — The two models are blended into a single output. A discretionary review then sits on top — covering gamestyles, scheduling, injury, and other qualitative factors the numbers cannot capture. Nothing is published without passing both.

3-Set Radar
7 criteria · Flag when 2+ trigger

The 3-Set Radar flags matches our model believes are likely to go the distance — all three sets in a best-of-three, or four or more sets in a Grand Slam. A match is flagged when two or more of seven criteria trigger simultaneously.


Skill Parity — how closely matched the two players are on the relevant surface.


Historical Pattern — how often their previous meetings have gone the full distance.


Form Convergence — whether both players are in similar form heading into the match.


3-Set Tendency (Main Tour) — each player's historical rate of going the distance at main tour level.


3-Set Tendency (Challenger) — the same measure at Challenger level, tracked separately.


Long Match Tendency — whether both players tend to play longer matches relative to the tour average.


H2H Tightness — whether their head-to-head record is competitive rather than one-sided.

How to use it
Useful for over/under sets markets and set betting. If a match is flagged, the market for "match to go to a deciding set" may be underpriced.
More criteria triggered = stronger signal. A match hitting 5 of 7 is a far stronger flag than 2 of 7. Weight your interest accordingly.
H2H Tightness and Historical Pattern together are particularly powerful — if they've consistently played long matches against each other, that pattern tends to repeat.
This is not a bet recommendation — it's a research signal. Always check the current market price before acting.
Serve & Return
Player stats — last 10 matches

This table shows two of the most important statistics in professional tennis for the players relevant to today's matches.


SPW% (Serve Points Won) — the percentage of points a player wins when they are serving. The tour average is around 63%. Players above this are strong servers. Players significantly below are vulnerable on serve.


RPW% (Return Points Won) — the percentage of points won when returning. The tour average is around 37%. Elite returners consistently push this above 40-42%.


The delta indicators (▲ / ▼) show two comparisons: how the player compares to the tour average, and how they compare to their own recent form. A player who is +7.8 vs tour but -3.4 vs themselves is performing above average overall but trending downward — a useful warning sign.


Numbers highlighted in yellow represent deltas of ±3 percentage points or more — a meaningful deviation worth noting.

How to use it
Look for mismatches. A player with very low RPW facing a player with very high SPW is a structural disadvantage — this kind of match is hard to win from the return.
The self-delta (vs own form) is often more useful than the tour delta. A player trending down in SPW is struggling on serve regardless of where they rank overall.
Surface matters — these stats are calculated per surface. A player's clay SPW can be very different to their hard court SPW.
Form
Last 10 matches W/L sequence

The Form section shows the last 10 match results for players in and around today's card, displayed as a W/L sequence from oldest to newest — left to right.


Reading the sequence left to right tells you whether a player is improving, declining, or inconsistent. A player whose sequence reads W W L W W W W W W W is on a strong run with one blip. A player reading W W W W L L L L L L is in a concerning collapse.


Best in Form shows the players with the most wins in their last 10, with notable scalps listed underneath. Worst in Form shows the opposite — players who are struggling, with the quality of opponents they've lost to noted for context.


A bad run against top-20 opponents is very different to a bad run against lower-ranked players. The opponent context matters.

How to use it
Don't just look at the record — read the sequence. 7-3 with the 3 losses at the start is very different to 7-3 with the 3 losses at the end.
Use form as a filter, not a signal. A player in terrible form facing a player in great form makes the model's edge number more credible. Form in the same direction as the pick adds conviction.
Be careful with hot streaks. A 10-0 player may simply have had a soft draw. Check who they beat.
A note on responsible use
The Compound publishes research and analysis. We are not a licensed financial advisor or gambling service. Nothing on this website constitutes financial or gambling advice. You are responsible for your own betting decisions.

Betting carries risk. Even high-edge picks lose. Over a sample of any size, variance is real. Only bet with money set aside specifically for this purpose, use sensible stake sizes, and if betting stops being enjoyable, stop.

If you are concerned about your gambling, visit begambleaware.org or call the National Gambling Helpline on 0808 8020 133.
Coming Soon
Football — How It Works