Variance vs edge

Edge is a real, repeatable advantage that makes money long-run; variance is the random swing of short-run results around that true rate. The trouble is they look identical over small samples: a lucky bettor with no edge and a skilled one can post the same record for months. Distinguishing them requires either a very large sample or a leading indicator like closing line value — which is why serious analysts trust CLV over win rate.

Flip enough coins and some streaks look like genius. Hundreds of bets sounds like a lot, but it isn't enough to separate a small real edge from noise — the confidence intervals are wide. That's why “I'm up over my last 200 picks” proves far less than people think, and why we don't treat a good stretch as proof of anything.

Our public ledger sits at 56.5% with +45.43u over 431 settled picks — a winning record. Yet our closing line value runs roughly flat (+0.39pp across 193 measured picks). Read together, that's the honest verdict: the profit is real but well within what variance can produce, and we have not demonstrated a durable edge over the closing market. We publish that rather than sell the win rate as proof. Anyone claiming a proven edge should show you their CLV, not just their record.

What is the difference between variance and edge?Edge is a real, repeatable advantage that profits long-run; variance is the random swing of short-run results around the true rate. Over small samples the two are nearly impossible to tell apart.

How many bets does it take to prove an edge?More than most people think — often thousands, because the confidence interval around a win rate is wide. A few hundred bets cannot reliably separate a small real edge from luck, which is why closing line value is used as a faster signal.

Does a winning record prove skill?No. A winning record over a few hundred bets is well within what variance alone can produce. Signal Labs' own 56.5% record with roughly flat CLV is a concrete example of profit that does not prove a durable edge.