WTDS FAV Report

HERMES ML PICK PERFORMANCE BY PRICE TIER A Report on Where We Actually Make (And Lose) Money

Based on 2,129 settled moneyline picks, 2026 season to date.

EXECUTIVE SUMMARY

After 2,129 settled moneyline picks across the 2026 season, a clear and actionable pattern has emerged. The edge is almost entirely concentrated in the underdog-and-small-favorite price zones. The medium-favorite price range between -140 and -199 has been a persistent drain on the P&L.

The single most important finding is this. Hermes has generated +$12,689 of profit on 712 picks in the +101 to +150 underdog bracket, for an ROI of +17.8%. This is not a small sample, not variance, not a fluke. It is by far the most profitable and highest-volume category of bet the system produces. Simultaneously, the system has lost nearly $7,200 combined on the -140 to -199 range across 483 picks. The numbers are stark enough that they should directly influence how picks are sized, and whether marginal picks in the losing ranges should be published at all.

THE FULL PICTURE

Raw data for all 2,129 settled moneyline picks by price range.

Heavy favorites priced -300 or worse went 7-15, 31.8% win rate, P&L -$1,299, ROI -59.1%. Favorites priced -250 to -299 went 28-15, 65.1% win rate, P&L -$435, ROI -10.1%. Favorites priced -200 to -249 went 38-13, 74.5% win rate, P&L +$438, ROI +8.6%. Favorites priced -170 to -199 went 88-71, 55.3% win rate, P&L -$2,254, ROI -14.2%. Favorites priced -140 to -169 went 166-158, 51.2% win rate, P&L -$4,870, ROI -15.0%. Favorites priced -110 to -139 went 311-227, 57.8% win rate, P&L +$3,141, ROI +5.8%. Pick'em games -109 to +100 went 157-151, 51.0% win rate, P&L +$216, ROI +0.7%. Small underdogs +101 to +150 went 374-338, 52.5% win rate, P&L +$12,689, ROI +17.8%. Medium underdogs +151 to +200 went 26-72, 26.5% win rate, P&L -$2,908, ROI -29.7%. Heavy underdogs +200 or better went 24-29, 45.3% win rate, P&L +$4,155, ROI +78.4%.

Totals come to 1,219 wins and 1,089 losses, net P&L +$8,873. Hermes is profitable overall, but almost entirely because of the underdog performance. The favorite side of the book is significantly underwater.

TIER 1: HEAVY FAVORITES, -200 AND DEEPER

The most surprising and statistically noisy area of the data. Three sub-brackets, each telling a different story.

The -300 and heavier bracket at 7-15 and -59.1% ROI matches intuition and external market research. When the public piles on a prohibitive favorite, books price them punitively. At -300 you need a 75% win rate just to break even. Hermes is getting 31.8%. The sample is small at 22 picks, but the directional signal aligns with documented market behavior. Heavy MLB favorites are systematically overpriced because public betting inflates them beyond their true probability.

The -250 to -299 bracket at 28-15 and -10.1% ROI is the quiet killer. The 65.1% win rate is excellent in absolute terms, but breakeven at those prices is 71 to 75 percent. Even 65% wins loses money over time. Picks that feel like they're winning, but bleeding slowly because the price is too high.

The -200 to -249 bracket at 38-13 and +8.6% ROI is the one exception in the heavy-favorite zone. Hermes is beating the implied probability with a reasonable 51-pick sample. Best guess on why. This bracket often represents genuinely dominant matchups where pricing still does not fully capture the edge. But it is fragile. 51 picks is not enormous, and one bad week could flip it negative.

Practical takeaway. Hermes should probably stop publishing picks on favorites priced -250 or worse.

Part 2 

TIER 2: THE MEDIUM FAVORITE TRAP, -140 TO -199

This is where Hermes bleeds the most money, and it's where tonight's Atlanta pick sits. Two sub-brackets combined cover 483 picks and a total loss of $7,124. By far the biggest single drain on the system's bottom line.

The -170 to -199 bracket went 88-71, 55.3% win rate, P&L -$2,254, ROI -14.2%. The -140 to -169 bracket is worse. 166-158, 51.2% win rate, P&L -$4,870, ROI -15.0%.

The uncomfortable truth is that Hermes actually picks winners in this range. 51 to 55 percent is not a bad win rate in absolute terms. The problem is breakeven at these prices is 58 to 67 percent, and the system is falling short by 5 to 7 percentage points. Over 483 picks, that margin compounds into real money lost.

Why does this happen. Best theory. These are games where Hermes sees enough signal to identify a favorite, but not enough to identify a true mispricing. The team has better pitching, better record, better power rating. But so do the books. The market already knows these teams are better, and the price reflects it. Hermes is co-signing the consensus, and books charge a juice premium for that consensus.

There is also a psychological component. These picks feel safest to bet. Mid-tier favorites look like value, like confident plays, and they encourage pressing. But the data says the opposite. The confident-looking mid-favorite is consistently the worst category the system produces.

Practical takeaway. Hermes should either stop publishing picks in the -140 to -199 range entirely, or require an unusually high cert score, say 85 or above, before publishing. For tonight specifically, Atlanta at -149 sits directly in the worst-performing bracket. The individual thesis is strong, but the bracket is a yellow flag. Size accordingly.

TIER 3: THE WORKING FAVORITE ZONE, -110 TO -139

Where the favorite side of the book starts earning its keep. 311-227, 57.8% win rate, P&L +$3,141, ROI +5.8%, across 538 picks. The largest favorite-side sample and clearly profitable.

Why this range works when -140 to -199 does not. The price is low enough that breakeven, between 52 and 58 percent, is easy to clear. Hermes's 57.8% win rate gives a real edge after juice. These are often games where both teams are close, and the favorite is only slightly ahead. Books have less public distortion to price against.

Picks in this range should be kept and can be sized at a normal unit.

TIER 4: THE SMALL DOG GOLD MINE, +101 TO +150

The headline finding of the entire report. 374-338, 52.5% win rate, P&L +$12,689, ROI +17.8%, across 712 picks. This single bracket is responsible for more than 100 percent of Hermes's net profit this season. Take it away and the rest of the book is basically breakeven or worse.

Why this range works. Small underdogs at plus money are systematically undervalued by the public market. The public heavily favors taking favorites, which pushes favorite prices up and underdog prices up with them. When Hermes correctly identifies an underdog that is actually closer to a coin flip than the odds suggest, even a modest 52.5% win rate produces a massive ROI because the payouts are larger than the risk.

The math is clean. At +120, breakeven is 45.5%. Hermes is winning 52.5%. That 7-point gap compounds across 712 picks. At +140, breakeven drops to 41.7%, and the gap is even wider. The system is exploiting a real and persistent market inefficiency.

Practical takeaway. Picks in the +101 to +150 range should be treated as the highest-priority publishable picks. A case can be made for upsizing them slightly above standard unit because the historical ROI is so strong. This is Hermes at its best.

TIER 5: THE MEDIUM DOG PITFALL, +151 TO +200

One of the more jarring results in the report. Medium underdogs, the bets that look juicy and contrarian, have been a disaster. 26-72, 26.5% win rate, P&L -$2,908, ROI -29.7%, across 98 picks.

This looks contradictory at first. If small dogs at +101 to +150 are profitable, why do dogs one tier up collapse. The answer is sharp selection bias. When a team is priced +101 to +150, the book sees them as legitimately competitive and the line reflects a close matchup. When a team moves to +151 or deeper, the book has made a real assessment that the team is significantly worse. Books are rarely that wrong. At +170 you need to win 37% of the time to break even. Hermes is winning 26.5%. The gap is enormous.

The 98-pick sample is medium-sized. Not enough to be totally confident the bracket is permanently dead, but enough that the pattern deserves serious respect. My read is that Hermes's contrarian-dog strategies in this range are identifying the wrong kind of underdog. The dog priced +170 usually deserves to be priced +170.

Practical takeaway. Picks +151 to +200 should probably be dropped until the system can identify what distinguishes a profitable medium dog from a bad one. Current performance is not variance. It is a real negative edge.

TIER 6: THE HEAVY DOG LOTTERY, +200 AND BEYOND

The most volatile category. 24-29, 45.3% win rate, P&L +$4,155, ROI +78.4%, across 53 picks.

The ROI is eye-popping, but the sample is small and the result is driven by a few big paydays. One or two wins at +400 swing the whole bracket positive. A cold streak could flip it.

What this tells us. Hermes is identifying genuine upset candidates at a rate that pays off when they hit. 45.3% on teams priced +200 or worse is substantially better than market expectation of 33% or less. That suggests a real signal, possibly tied to situational spots like rested teams hosting exhausted travelers.

Practical takeaway. Keep publishing heavy dogs, but acknowledge the variance. Size at a smaller unit than the +101 to +150 bracket. These are lottery tickets with positive expected value.

CONCLUSIONS AND RECOMMENDATIONS

The data tells a consistent story. Hermes's edge is on the underdog side, particularly the +101 to +150 bracket. The favorite side is roughly breakeven at -110 to -139 and a drain elsewhere, especially in the -140 to -199 range where 483 picks produced a loss of $7,124.

Five concrete recommendations.

First. Drop or drastically filter the -140 to -199 bracket. An ROI of -14% to -15% across 483 picks is not survivable. Either skip these, require cert 85 plus, or publish with a warning.

Second. Drop -250 and worse entirely. Win rates look fine but prices make them unprofitable.

Third. Upsize the +101 to +150 bracket. This is the most profitable category and deserves priority sizing. A full unit or 1.25 units rather than a partial.

Fourth. Drop +151 to +200 until further analysis. Current -29.7% ROI is a structural problem.

Fifth. Keep +200 plus heavy dogs but size conservatively. Strong signal, high variance.

If these recommendations had been applied all season, Hermes's net P&L would likely be in the +$15,000 to +$18,000 range instead of +$8,873. The system has a real edge. It just needs to stop giving back what it earns.

Simple version:

THE RAW DATA

Across 270 games where the favorite was priced between -110 and worse, here is what happened when you bet the favorite $100 flat at the best available pre-game line.

Heavy favorites priced -300 or worse went 4-2, a 66.7% win rate, P&L -$141.29, ROI -23.55%.

Favorites priced -250 to -299 went 5-6, a 45.5% win rate, P&L -$414.18, ROI -37.65%.

Favorites priced -200 to -249 went 8-0, a 100% win rate, P&L +$359.42, ROI +44.93%.

Favorites priced -170 to -199 went 17-7, a 70.8% win rate, P&L +$247.55, ROI +10.31%.

Favorites priced -140 to -169 went 33-45, a 42.3% win rate, P&L -$2,334.55, ROI -29.93%.

Favorites priced -110 to -139 went 75-68, a 52.4% win rate, P&L -$663.80, ROI -4.64%.

Total across all tiers: 270 bets, net P&L -$2,946.84, ROI -10.91%.

Thanks very much for posting this. 

Tons of info