Yield (ROI) in Value Betting
Published Thu Sep 04 2025
Yield (often called ROI) measures efficiency of your staked capital.
Yield % = (Total Profit / Total Staked) * 100If you staked 10,000 units across bets and profited 600 units:
Yield = 600 / 10,000 = 6%If average stake is 1 unit, that’s 600 units profit over 10,000 units risked—simple but powerful.
Yield vs EV vs CLV
| Metric | What it reflects | Time Sensitivity |
|---|---|---|
| EV (expected) | Model / market probability edge | Immediate (pre-bet) |
| CLV | Execution & price capture quality | Near-term (at start) |
| Yield | Realized profitability | Lagging (needs volume) |
Healthy process: Positive EV & positive CLV eventually produce positive yield.
Pitfalls
- Small samples exaggerate yield.
- High yield from a single large underdog skewing results.
- Ignoring CLV: yield without supporting process metrics may be luck.
- Forgetting stakes: reporting ROI without context of variance & odds profile.
- Survivorship bias (dropping dead strategies from aggregate yield chart).
Normalize & Segment
Track yield by market type, sport, odds range. This reveals where edge is real vs variance.
Example Segmentation Table
| Segment | Bets | Avg Odds | Mean Stake | Yield % | CLV % |
|---|---|---|---|---|---|
| Soccer Major | 450 | 1.95 | 1.00 | 3.2 | 1.1 |
| Soccer Lower | 220 | 2.12 | 0.80 | 7.5 | 2.4 |
| Tennis | 180 | 1.75 | 1.10 | -1.0 | -0.2 |
| Basketball | 300 | 1. ninety? | 1.00 | 5.0 | 1.3 |
Tennis segment shows negative yield & near zero CLV—candidate for review or refinement.
Expected vs Observed
Your long-run actual yield should converge toward your average EV per bet minus friction (limits, vig drift, model error).
Statistical Significance (Simplified)
Approximate standard error (SE) of yield for even-money style bets:
SE ≈ sqrt(Variance of unit returns / Total Staked)If unit return variance ≈ 1 and you’ve staked 10,000 units:
SE ≈ sqrt(1 / 10,000) = 1%A 6% yield is ~6 SE above zero → likely real. At only 500 units staked:
SE ≈ sqrt(1 / 500) ≈ 4.47% -> 6% is ~1.34 SE (could easily be noise)Adjusted Yield
Remove outlier windfalls (huge longshot wins) and recompute “core yield” to evaluate stability.
Risk-Adjusted Perspective
Two bettors each have 6% yield. One had maximum drawdown 12%, the other 40%. The smoother equity curve indicates better risk management. Track Sharpe-like ratio:
Risk Adjusted Return = (Average Unit Return) / (Std Dev Unit Return)FAQs
How many bets before yield is meaningful? Depends on variance & edge size. For modest edges at average odds 2.00, thousands of bets give stability.
Can yield decline even if my CLV improves? Yes short-term; realized results lag process improvement.
Should I chase higher yield segments aggressively? Confirm capacity: increasing volume might erode price quality (deteriorating CLV).
Why is my yield lower than average EV? Model error, line movement after placement, or sampling variance; review calibration & execution speed.
Takeaway
Yield is the scoreboard. Combine it with EV estimates and CLV tracking for a three-dimensional performance view.