SharpSDA
Help / What the Flames Mean

What the Flames Mean

The rating is a deterministic summary of how much edgeThe gap between the model's win probability and the market-implied probability at bet time. Positive edge means the model thinks the outcome is more likely than the price implies. the model sees on the picked side. It is computed at publish and never changes โ€” you can compare picks across different dates on an apples-to-apples basis.

Rating bands

RatingPicked-side edgeMeaning
๐Ÿ”ฅ3โ€“4%Mild edge. Bet small or skip if the line moves.
๐Ÿ”ฅ๐Ÿ”ฅ4โ€“5%Solid edge. Standard Kelly sizing.
๐Ÿ”ฅ๐Ÿ”ฅ๐Ÿ”ฅ5โ€“8%Strong edge. Full recommended stake.
๐Ÿ”ฅ๐Ÿ”ฅ๐Ÿ”ฅ๐Ÿ”ฅโ‰ฅ 8%High-conviction. Prioritize getting this line.
MUST BETโ‰ฅ 10%Edge โ‰ฅ10% AND model prob โ‰ฅ60% AND Kelly โ‰ฅ2u. Highest tier.

What edge means here

Picked-side edge = the model's calibrated win probabilityThe model's calibrated probability that the picked side wins. This is distinct from the market-implied probability โ€” the difference between the two is the edge. for the picked team minus the implied probabilityThe win probability encoded in a betting line. Convert American odds: negative odds โ†’ |odds|/(|odds|+100); positive odds โ†’ 100/(odds+100). E.g. -150 โ‰ˆ 60%, +130 โ‰ˆ 43%. from that team's published odds. It answers the question: how much better does the model think this team's chances are compared to what the market is pricing?

What rating is NOT

The rating does not predict that a 4๐Ÿ”ฅ pick will win more often than a 2๐Ÿ”ฅ pick in any single game. Baseball is random in the short run. The rating says the model found a larger pricing discrepancy โ€” over many picks, higher-rated picks should produce better ROIReturn on Investment. Net profit divided by total units staked. Measures efficiency across bets of different sizes. and more positive CLVClosing Line Value. The difference between the close-price implied probability and your bet-price implied probability. Positive CLV means you got a better price than where the market settled โ€” the best predictor of long-run edge..

The Results page shows track record broken down by rating band so you can verify this relationship in actual settled picks.