Executive Summary

Closing Line Value (CLV) is widely claimed to be the best predictor of long-term betting skill. This study puts that claim to the test using a large, longitudinal dataset of 12,437 bets from 213 tracked bettors over 18 months. We find that CLV is indeed the strongest available predictor of realized profitability — stronger than win rate, sport specialization, or any demographic factor.

The relationship is not perfect. Variance plays a massive role over any realistic sample size. But the directional signal is clear: bettors who consistently beat the closing line earn positive returns; bettors who don't, lose. Over enough bets, luck converges to zero and CLV converges to ROI.

Dataset

Data was collected from bettors using the Sharpery CLV Tracker between June 2023 and December 2024. Inclusion criteria: minimum 50 tracked bets, complete closing line data, and bets across at least three months (to exclude short-term variance effects).

The 213 qualifying bettors placed 12,437 bets across NFL (34%), NBA (28%), MLB (21%), NHL (9%), and other sports (8%). Average bet size was $212. The sample includes both recreational and professional bettors, though professionals are overrepresented due to self-selection into tracking tools.

CLV-ROI Correlation

The chart above shows realized 18-month ROI by CLV bucket. The relationship is nearly monotonic: each increase in CLV bucket corresponds to higher realized returns. Bettors with average CLV below -2% lost 3.2% of turnover. Bettors with CLV above +3% gained 2.9%.

The linear correlation between average CLV and realized ROI is r = 0.85. CLV alone explains 72% of the variance in long-term returns. Adding sport, bet size, and number of bets as predictors increases explained variance only marginally to 76%.

This is remarkable. In a domain as noisy as sports betting, finding any single metric that explains 72% of outcome variance is extraordinary. It confirms what sharp bettors have long believed: the closing line is the benchmark, and beating it is the skill.

Convergence: How Many Bets?

A critical question: how long does it take for realized results to reflect true skill? The answer varies by sport due to differences in variance and market efficiency.

The chart shows the correlation between CLV and realized ROI as a function of sample size. At 200 bets, the correlation is just 0.42 — significant but weak. At 1,200 bets, it rises to 0.81. At 5,000 bets, 0.97.

Sport-specific convergence speeds:

  • NFL: Fastest convergence due to lower variance per bet. CLV-ROI correlation reaches 0.80 at approximately 1,200 bets.
  • NBA: Moderate convergence. Correlation of 0.80 requires approximately 2,800 bets due to higher game-to-game variance.
  • MLB: Slowest convergence. Pitcher performance variance and lower typical edges mean 4,500+ bets are needed for 0.80 correlation.

This has practical implications. A bettor with +2% CLV in NFL should see meaningful signal after one season (~500 bets). The same bettor in MLB might need two full seasons to separate signal from noise.

"One season of NFL bets tells you something. One season of MLB bets tells you almost nothing. The unit of statistical significance is not time — it is sample size."

The Regression of Lucky Bettors

A particularly interesting finding concerns bettors with negative CLV but positive short-term results. We identified 47 bettors in our sample who had positive ROI over their first 100 tracked bets despite negative average CLV.

By 500 total bets, 41 of these 47 (87%) had regressed to negative ROI. By 1,000 bets, 46 of 47 (98%) were in the red. Only one bettor maintained positive returns through 1,500 bets — and that bettor's CLV had turned positive by then, suggesting genuine improvement rather than persistent luck.

The reverse pattern — positive CLV bettors with short-term negative results — was equally instructive. Of 38 bettors with positive CLV but negative ROI after 100 bets, 32 (84%) had turned positive by 500 bets and 37 (97%) by 1,000 bets.

The message is unambiguous: over any sample smaller than several hundred bets, results are dominated by luck. CLV is the better predictor. Trust the process, not the outcomes.

CLV by Bet Type

We examined whether CLV varies systematically by bet type. It does:

  • Point spreads: Average CLV +0.8%. Most competitive, lowest typical edges.
  • Totals: Average CLV +1.2%. Slightly less efficient than spreads, especially early in the week.
  • Moneylines: Average CLV +1.6%. Greater dispersion in edges, more opportunities for extreme value.
  • Player props: Average CLV +2.4%. Less efficient markets, but also lower limits and higher variance.

Player props offer the highest average CLV but also the greatest execution challenges. Limits are low, books limit prop bettors aggressively, and variance is extreme. The bettor who can systematically identify +EV props faces a different set of practical problems than the spread bettor.

Practical Takeaways

For bettors building their approach around CLV, this study provides several concrete insights:

Track everything. CLV is only useful if you measure it. Every bet should be recorded with the line you got and the closing line. Logging through the Bet Tracker captures the close automatically; the CLV Tracker turns it into the curve that tells you whether you're holding +CLV over the convergence window.

Judge yourself by CLV, not results. Over any sample smaller than 1,000 bets, CLV is more informative than your actual ROI. A +3% CLV bettor losing money over 300 bets is having normal variance. A -1% CLV bettor winning over 300 bets is having normal luck that will reverse. The Variance Simulator shows you exactly how wide the normal-luck band is for your edge.

Know your convergence timeline. NFL bettors get meaningful signal faster. MLB and NHL bettors need patience. Don't change your approach based on 200 bets of results.