CLV is the only metric that survives variance — how to track it without lying to yourself
Win-loss tells you nothing useful at 100 bets. CLV tells you whether the market eventually agreed with you. The cleanest signal of edge available in short timeframes — if you measure it honestly.
- CLV
- Methodology
Over 100 bets, win-loss P/L is mostly variance — a strategy with no real edge can show +12% ROI, a strategy with a real 2% edge can show -8% ROI, and you can’t tell the difference. CLV strips variance out: it measures whether the market eventually agreed with you, which converges much faster than P/L. Sharp operators track CLV per strategy and ignore W-L over short samples. Glitch Edge ships CLV telemetry per strategy in Free and Pro.
Why P/L lies in short samples
The math is unforgiving. For a strategy with a real 2% edge betting at average decimal odds 2.00, the standard deviation of P/L per bet is roughly the bet size. After 100 bets at $100 stake, the standard error around expected P/L of +$200 is roughly $1,000 — meaning a 95% confidence interval spans -$1,800 to +$2,200.
In plain terms: a strategy with a real +2% edge can show anywhere from -18% to +22% ROI over its first 100 bets, all due to variance. You can’t tell from inside the run whether the edge is real, decayed, or never existed.
What CLV measures instead
For a single bet at decimal odds o_taken with closing decimal odds o_close:
CLV % = (o_taken − o_close) / o_close × 100
Positive CLV means you got better odds than the close — the market eventually agreed with your direction. Negative CLV means the line moved against you. Aggregated across bets, average CLV is the cleanest available signal of whether you’re consistently beating the close.
CLV variance is much lower than P/L variance because closing lines are a relatively stable price-discovery mechanism. The same 100-bet sample that gives you a wild ±20% P/L confidence interval gives you a CLV confidence interval roughly 3× tighter.
The CLV thresholds worth knowing
After 100+ tracked bets:
| Aggregate CLV | Verdict |
|---|---|
| > +2% | Strong edge. Strategy is consistently beating the close. |
| +0.5% to +2% | Real edge — sustainable if maintained. |
| -0.5% to +0.5% | Flat. Near break-even against the market. |
| < -0.5% | No edge at this sample size; strategy needs work or retirement. |
The thresholds are different for different markets. Sharp markets (Pinnacle, Cloudbet sharp lines) move efficiently, so beating them by even +1% is a real edge. Recreational markets move sluggishly, so +2% is roughly the floor for a “real” sharp edge in those markets.
How to track CLV without lying to yourself
The trap is selecting which bets you measure. Five honesty rules:
- Track every placed bet, not just winners. Cherry-picking which bets to include in the CLV calculation guarantees positive CLV.
- Use the actual close, not pre-game lines. The closing line is the final aggregate guess at true probability. Pre-game lines from 4 hours before close haven’t incorporated late information.
- Compare to the sharp close, not the book you placed at. If you placed at recreational books that don’t move much, comparing to their own close understates CLV. Use Pinnacle or Cloudbet’s close as the reference.
- Don’t strip vig manually before computing. CLV is computed on the offered odds, not on no-vig adjusted prices. The vig is part of the bet you took.
- Aggregate per strategy, not portfolio-wide. Portfolio CLV averages can hide one strategy carrying the program while three others lose money.
Tracking CLV in Glitch Edge
The platform records the offered odds and the closing odds for every placed bet automatically. The weekly strategy report shows:
- Per-strategy aggregate CLV
- Per-recipe breakdown within each strategy
- Trend over time (rolling 30-day CLV vs lifetime CLV)
- Beat-the-close ratio (% of bets where you got better odds than close)
- Promotion-readiness flag for strategies in paper mode (need 50+ bets + positive CLV before promotion to live)
This is what most off-the-shelf betting tools don’t ship — the closing-line snapshot at the exact bet’s resolution time, fed into a per-strategy aggregator.
CLV is NOT a guarantee of profit
The honest framing matters. CLV signals that the market eventually agreed with you. It does not guarantee future profit. Specifically:
- CLV can decay. A strategy with +1.5% CLV for 6 months can decay to +0.2% CLV as the market learns the inefficiency. Track rolling CLV, not just lifetime.
- CLV doesn’t account for limiting. Sharp accounts get limited at recreational books. A strategy with positive CLV that’s getting account-limited isn’t sustainable regardless of the math.
- CLV at recreational books is softer. A strategy can post +2% CLV at recreational books and -0.5% at sharp books, because the recreational close didn’t move efficiently. Treat recreational-book CLV with more skepticism.
When P/L still matters
CLV is the short-term signal. Over thousands of bets, actual P/L is what pays the rent. The relationship between CLV and P/L is well-established but noisy: a strategy with consistent +1% CLV typically realizes roughly +1% ROI over 2,000+ bets, with substantial variance around that.
So:
- Track CLV in the short term (1–500 bets) to know whether the edge is real
- Track P/L in the long term (1,000+ bets) to know what you actually made
Both metrics matter; they just measure different things.
Frequently asked questions
How is CLV different from expected value (EV)?
EV is forward-looking — the probability-weighted return of a bet at decision time. CLV is backward-looking — how much better than the eventual close you got. Both are signals of edge; CLV converges faster.
Can I compute CLV after the fact for bets I placed elsewhere?
Yes — use the CLV Tracker. Paste your bets as CSV (odds taken, closing odds, stake) and get aggregate CLV per row plus a verdict.
What about in-play bets where there isn’t a “close”?
In-play CLV is harder — there’s no single closing line. The standard approach is to compare to the line 5–10 seconds after your placement; if the line moved in your direction, that’s positive intra-play CLV.
Does Glitch Edge use CLV to gate promotion to live placement?
Yes. The default promotion rule is 50+ paper bets with positive aggregate CLV. The threshold is configurable per strategy.
Why not just track win rate vs implied probability?
That works but is noisier than CLV. Win rate variance scales with sample size; CLV variance scales with price-discovery noise (much lower). For the same sample size, CLV is the cleaner signal.