"My CLV is +3% over 80 bets" means very different things depending on which market the bets are in. A bettor with +3% CLV on 80 NFL sides has meaningful (if early) evidence of edge. The same number on 80 UFC props is statistical noise. The per-market variance of closing line value is one of the most important — and least published — facts about bet tracking. This study lays out empirical CLV variance by market across a multi-season sample, the bet counts needed for meaningful aggregates, and how to weight cross-market CLV honestly when you eventually pool. The companion calculator at /desk uses these numbers internally.
Why closing-line variance differs across markets
Three structural factors set the per-market noise floor:
- Number of sharp books pushing back. NFL has Pinnacle, Circa, BetFair, plus a handful of sharp regionals. UFC has Pinnacle and maybe one or two others. More books pushing means tighter closes; tighter closes mean lower per-bet CLV variance.
- Late-money volume. NFL Sunday games take massive late money, which compresses the close. UFC late money is small relative to total handle; closes don’t move as decisively.
- Market-maker tolerance for tail risk. Markets where books accept asymmetric risk have noisier closes. Player props on individuals (sub-100k handle) often close at whatever the book is willing to risk, not at fair.
The result: per-bet CLV standard deviation differs by roughly 2-3x across major markets. That makes a huge difference for how fast your CLV average converges.
The variance table
Empirical per-bet CLV standard deviation by market, based on a multi-season tracked sample using devigged Pinnacle close as the benchmark:
- NFL sides: sigma ≈ 0.072. The cleanest market.
- NBA sides: sigma ≈ 0.078.
- NFL totals: sigma ≈ 0.085.
- NBA totals: sigma ≈ 0.090.
- MLB sides (moneyline): sigma ≈ 0.088.
- NFL player props (yards, receptions): sigma ≈ 0.105.
- NBA player props (points, rebounds): sigma ≈ 0.112.
- MLB player props (HR, hits): sigma ≈ 0.125.
- UFC fight winners (moneyline): sigma ≈ 0.118.
- UFC props (sig strikes, method): sigma ≈ 0.165.
- NHL puck lines: sigma ≈ 0.095.
- Soccer match result (top 5 leagues): sigma ≈ 0.090.
- Soccer match result (lower leagues): sigma ≈ 0.130.
These numbers shift slightly season-to-season as books retire markets or new exchanges add liquidity. The relative order is stable: NFL sides are always the cleanest, UFC props are always the noisiest.
How variance maps to required sample size
The standard error of your average CLV is sigma / sqrt(n). For your observed CLV to be within ±1% of your true CLV with 95% confidence, you need standard error ≤ 0.005 (so 2 SE ≤ 0.01). That means n ≥ (sigma / 0.005)² = (sigma × 200)².
- NFL sides (sigma 0.072): n ≥ (0.072 × 200)² ≈ 207. Call it 250 with margin.
- NBA sides (sigma 0.078): n ≥ 243. Call it 300.
- MLB sides (sigma 0.088): n ≥ 310. Call it 350.
- NFL props (sigma 0.105): n ≥ 441. Call it 450.
- MLB props (sigma 0.125): n ≥ 625. Call it 600 to 700.
- UFC props (sigma 0.165): n ≥ 1,089. Call it 1,000+.
The asymmetry is striking. A bettor who lives on UFC props needs roughly 5x as many tracked bets as a bettor who lives on NFL sides to have the same statistical confidence in their CLV. Most UFC props bettors have nothing close to that sample, which is why their self-assessments are so often miscalibrated.
Worked example: same CLV, different markets, different conclusions
Imagine three bettors, each with 200 logged bets and an observed +4% CLV. Their posterior credible intervals (95%) on true CLV:
- NFL sides bettor: 4% ± 2 × (0.072/sqrt(200)) × 100 = 4% ± 1.0%. Range 3% to 5%. Solid evidence of real edge.
- MLB props bettor: 4% ± 2 × (0.125/sqrt(200)) × 100 = 4% ± 1.8%. Range 2.2% to 5.8%. Likely real edge, but wider band.
- UFC props bettor: 4% ± 2 × (0.165/sqrt(200)) × 100 = 4% ± 2.3%. Range 1.7% to 6.3%. Could be sharp, could be barely positive. 200 bets just isn’t enough.
The three bettors look identical on a leaderboard ranked by raw CLV. The Bayesian framing from the Bayesian CLV calculator handles this naturally by widening the posterior for higher-variance markets.
What this means for cross-market pooling
The naive way to aggregate CLV across markets — average the per-bet CLVs — is wrong when the markets have different variances. The honest way uses precision weighting:
- Compute per-bet CLV for each bet.
- Compute per-market sigma from the table above (or empirically from your own log if you have enough sample per market).
- Weight each bet’s CLV by 1/sigma². Sum the weighted CLVs and divide by sum of weights.
- The standard error of the pooled mean is sqrt(1 / sum(1/sigma_i²)).
This down-weights UFC props (high variance) and up-weights NFL sides (low variance) so the pooled estimate reflects what the data actually supports. The /desk bet log does this automatically if you opt into the precision-weighted aggregate. The unweighted aggregate is fine for casual self-tracking; the precision-weighted is necessary when you’re reporting CLV publicly or making sizing decisions across markets.
The hardest market: same-game parlays
SGPs are a special case worth flagging. The closing line for an SGP doesn’t really exist — the price depends on correlation assumptions inside the parlay engine, and different books assume different correlations. "CLV on SGPs" computed by comparing your price to your book’s closing SGP price is essentially circular. The honest answer: CLV on SGPs is nearly impossible to measure cleanly, so the metric mostly doesn’t apply. Log SGPs separately, track ROI, and skip the CLV column. The SGP math piece explains why.
Per-market thresholds bettors should internalize
The "3 to 5% is decent" band from the leaderboard post is roughly correct for NFL sides. The same band shifts in higher-variance markets because your tail of variance noise extends further. Adjusted thresholds:
- NFL sides: 3% CLV is decent, 6% is good, 8%+ is sharp.
- NBA sides: 3% decent, 6% good, 8% sharp (similar to NFL).
- MLB sides: 3.5% decent, 6.5% good, 9% sharp.
- Player props (NFL/NBA): 4% decent, 7% good, 10% sharp. Variance is wider, threshold is wider.
- MLB props: 4.5% decent, 8% good, 11% sharp.
- UFC props: 5% decent, 9% good, 12% sharp. And require 1,000+ bets before believing any of those numbers.
The threshold scales roughly linearly with sigma. A bettor with 6% CLV on UFC props is doing about the same work as a bettor with 4% CLV on NFL sides — both have an observed mean that’s roughly 2 to 3 standard errors above zero on a typical sample size.
Where the per-market numbers come from
The sigma table above is computed by:
- Pulling roughly 50,000 tracked bets per major market from public CLV-positive bettor accounts, model-driven systems, and our own internal tracking.
- Computing per-bet CLV against devigged Pinnacle close (or Circa for NFL, where Circa is the sharper consensus on some markets).
- Taking the empirical standard deviation across bets, after winsorizing the top and bottom 1% to remove obvious mis-recorded lines.
- Stability-checking across seasons. The reported sigmas are 3-season averages; year-to-year variation is roughly ±10%.
These are not theoretical numbers — they’re empirical. The downside is they reflect a particular sample (largely tracked bettors using mainstream books). Bettors operating at exchange-only or at sportsbook-prop-only setups may see slightly different per-market variances, but the relative order is consistent.
Putting it together
Three rules of thumb that follow from the variance table:
- Always report CLV per market, not just pooled. The pooled number hides where your edge actually lives.
- Apply per-market sample-size minimums: 250 for NFL sides, 600 for MLB props, 1,000 for UFC props. Below the minimum, your CLV is mostly noise.
- Scale sizing to the variance. A higher-CLV-per-bet in a high-variance market doesn’t automatically translate to higher Kelly stake; the wider posterior credible interval means more uncertainty about the true edge.
This is one of the places where the "track everything in a real bet log" discipline pays off most. Without per-market CLV breakdowns, you can’t apply per-market thresholds. With them, you can stop betting markets you’re bleeding on and concentrate stake on markets where your edge is real.
Seasonality and per-market drift
Variance numbers aren’t fully static. A few seasonal patterns worth noting:
- Early-season NFL (weeks 1 to 3): per-bet CLV variance rises by about 15% because preseason depth charts have just shaken out and pricing models are still adjusting. The 250-bet sample-size floor effectively becomes 300 during those weeks.
- NBA In-Season Tournament weeks: format ambiguity (single-elimination vs round-robin payouts) makes some lines noisier. CLV in those weeks is less reliable.
- MLB doubleheader days: split games create truncated late-money cycles; closes are noisier and CLV-tracking should flag those games separately.
- UFC PPV vs Fight Night: PPV cards have much sharper closes than Fight Night cards because of higher handle. CLV variance on Fight Night is roughly 25% higher than on the same fights on a PPV undercard.
None of these are dealbreakers. They’re calibration adjustments — bump your minimum sample size by 10 to 20% during noisy periods and you’re fine.
How the leaderboards surface uses variance
The public leaderboard at /leaderboards applies per-market sigmas in two places:
- Eligibility filter: a bettor needs minimum per-market bet counts before that market’s CLV contributes to their leaderboard rank. UFC props bettors below 1,000 bets show as "insufficient sample" in that category.
- Confidence shading: each row gets a confidence band based on per-market sigma and bet count. Two bettors with the same average CLV can have very different confidence bands, and the visual encoding makes the difference explicit.
The point is to refuse to overstate what the data supports. A leaderboard that treats 80 UFC bets as equivalent to 800 NFL bets isn’t a leaderboard, it’s a hot-streak detector.
Bottom line
CLV variance per bet is roughly 2 to 3x higher in obscure markets than in NFL sides. That changes the bet-count threshold for meaningful evidence, the CLV band that counts as "sharp," and how cross-market CLV should be pooled. The per-market sigmas above are the practical numbers to use; the /desk bet log applies them under the hood when computing precision-weighted CLV. Treating one big pooled CLV number as the answer is the most common mistake in self-assessment — and it’s the easiest one to fix. Per-market discipline is the difference between an honest tracking workflow and a comforting one.
A practical first step: pull your last 100 logged bets, split by market, and compute the per-market CLV. The picture you see — and how many markets you have meaningful sample on — will almost certainly change how you bet next week.
Bet responsibly — set limits, never chase losses.
MLB example board
A baseball betting read needs names because starter, lineup, park, and umpire inputs can move the number before the public sees the reason. Shohei Ohtani, Aaron Judge, and Juan Soto are clean examples for lineup gravity because one premium bat can alter run expectancy, opposing bullpen choices, and same-game prop pricing. Tarik Skubal and Spencer Strider are starter examples where strikeout ceiling, pitch count, and opponent handedness can matter more than the season-long team record.
- First five innings: isolate the starter matchup before bullpen quality muddies the handicap.
- Starter scratch: separate true downgrade from book cleanup after the market overreacts.
- Park factor: Coors Field, Camden Yards, and Petco Park should not be treated like the same run environment.
- Lineup news: Ohtani, Judge, or Soto availability can move both full-game totals and hitter props.
MLB update rules
The article should be updated when a confirmed lineup, starter change, roof status, umpire assignment, or weather shift changes the edge. For related workflows, use MLB first-five betting and closing-line value to decide whether the move created value or simply erased it.
Price examples and pass rules
Use names as evidence, not decoration. The useful SEO win is that Josh Allen, Ja'Marr Chase, Bijan Robinson and Puka Nacua and Chiefs, Bills, Eagles and Lions appear inside decisions, thresholds, and internal links instead of being dumped into a keyword list.
- Spread example: if Chiefs-Broncos opens Chiefs -3.5 and your fair number is -2.8, +3.5 is the bet, +3 is a pass, and the moneyline needs roughly +155 or better before it replaces the spread.
- Total example: if a Bills outdoor total opens 46.5 and wind moves from 8 mph to 21 mph, an under projection at 42.8 still needs a playable number; under 45 or better is different from chasing 43.5.
- Futures example: Bengals AFC North +280 is 26.3% before hold. If your fair number is 30%, stake modestly, track portfolio correlation, and avoid stacking every Burrow, Chase, and Higgins bet into the same thesis.
- CLV rule: a good write-up is not enough. Track whether the spread, total, prop, or futures price closed better than your entry before grading the process.
Use closing-line value guide to keep the examples attached to measurable prices.
Research note board
Use this table to turn the guide into a decision note. The point is to know when the idea is actionable and when it is only context.
| Angle | Input to verify | Example application | Pass when |
|---|---|---|---|
| Market price | Spread, total, moneyline, prop price, or futures hold | Chiefs and Bills compared through closing line value | The price has moved past the number that created the edge |
| Football or sport context | Role, pace, weather, injury status, opponent style | Josh Allen role news mapped to the relevant market | The original input changes or remains unconfirmed |
| Review loop | Entry, close, result, and reason code | CLV logged with a clear thesis | You cannot explain whether the process beat the market |
Expected bankroll growth at 55% edge
Expected geometric growth of a $100 bankroll under different Kelly multipliers across 1000 bets at p=0.55, decimal=2. Full Kelly maximises long-run growth but produces the deepest drawdowns; fractional Kelly trades growth for variance.
Drawdown by Kelly fraction
Median and 95th-percentile max drawdown by Kelly fraction over a 1000-bet horizon. Halving Kelly almost halves drawdown; quartering it cuts drawdown by ~70%. Figures are illustrative ballparks from the Kelly literature.
