MONTHLY CLOSE · TRANSPARENCY · JULY 2026
July: a positive, inconsistent and decisive month for the algorithm
80 Premium Picks, four versions of the protection strategy and three families of Deep Signals. This is the complete balance: what won, what lost, how much risk it took on and what the month taught.
July ends with an official profit in Premium, but it would be misleading to sum it up with just one green number. The curve had a significant drawdown, performance varied a lot between surfaces and tours, and the moneyline protection was the difference between closing the month positive or negative.
What each figure means
The Premium result exactly follows the execution the user received on each date: main stake, odds and protection when it existed. ROI divides the profit by all the units actually exposed. Drawdown is the largest drop from a previous peak of the curve, not the sum of all losses.
Deep Signals are measured at 1u flat per signal on the match-winner market. Winner Elite, Tapado IA and Combi Elite are presented separately. When an aggregate with the combos is offered, each system is treated as an independent entry; that's why a Combi's legs can coincide with individual signals.
Premium: July's official result
A theoretical 100-unit bankroll that had followed every official execution would have ended at 108,39u, an 8.39% gain over the starting bankroll. It isn't a repeatable monthly interest rate or a forecast: it simply translates the month's units into the bankroll reference used in the project.
The opening days concentrated a significant part of the monthly profit.
The curve afterward was inconsistent and reached a maximum DD of −23.38u.
By calendar blocks, days 1–7 contributed +6.12u and days 8–14, +7.26u. The second half was weaker: from the 15th to the 21st, 3.13u were lost, and from the 22nd to the 31st, another 1.86u. The month ended positive thanks to the margin built in the first fortnight and the protection.
The protection wasn't decorative
During July there were 40 Premium Picks with a second moneyline entry. The result of the two legs makes it clear what each one did:
Without that second leg, the main block would have closed July at −2.72u. With protection, the official result was +8.39u. There were also 10 protected matches: the main scenario failed —for example, winning 2-0— but the read on who would win the match was correct.
It's worth interpreting the term “protection” correctly. It doesn't eliminate risk: it adds exposure. When the chosen player loses the match, both entries can fail. Its usefulness depends on the moneyline odds and the quality of the read compensating for that extra risk.
How the algorithm evolved into V4
July was also the month when a recurring read became a formal strategy. Many −1.5 sets picks failed because the predicted player won 2-1. The response wasn't to rewrite the pick, but to turn the main scenario and its protection into a single auditable Premium position.
- V1 · July 16: the −1.5 sets keeps its stake and adds the moneyline. With winner confidence of 82 or higher, protection could reach
min(2 × stake, 6u); below that, it stayed at 1u. - V2 · July 17: the same logic extends to the exact 2-0 score in BO3, because it represents the same sporting scenario.
- V3 · July 18: caps the main entry at 3u and sets protection at twice the main entry. It reduced some extremes, but could still push exposure too high.
- V4 · July 27: assigns 4u if the main odds are below 1.85; 3u between 1.85 and 1.89; and 2u from 1.90 up. It only adds 4u of protection when winner confidence is at least 75 and its odds are 1.25 or higher.
What has happened since V4's real activation
V4's live sample is still small: eight picks, five main entries correct and three failed. The main entry contributed +4.24u, protection subtracted −6.28u, and the block ended at −2.04u over 53u, ROI −3.85%. Eight cases don't justify changing a formula that was just frozen.
The historical check since the start of 10.8
Applying V4 retrospectively to the 77 eligible and settled Premium BO3 since the start of 10.8 would have produced +81.36u over 468u, 17.38% ROI and a maximum drawdown of −12u. The main entry would contribute +39.56u and protection, +41.80u.
It's a simulation, not the result actually achieved. Its purpose is to justify that the rule deserves prospective validation. The real test of V4 begins with the matches after July 27.
Where it won and where it lost
By surface
- Hard court: 36 picks, 22W–12L–2V, +13.56u and 6.58% ROI.
- Grass: 16 picks, 11W–5L, +12.70u and 19.24% ROI.
- Clay: 28 picks, 9W–18L–1V, −17.87u and −12.95% ROI.
Clay was Premium's biggest problem. Within the full 10.8 audit, the −1.5 sets market on clay built up a sample negative enough to justify a specific prospective control. That doesn't make all clay bad: the problem lies in combining surface and market.
By tour
- Challenger: 29 picks, 20W–7L–2V, +37.60u and 25.75% ROI.
- ATP: 39 picks, 15W–23L–1V, −28.63u and −13.50% ROI.
- Wimbledon: 12 picks, 7W–5L, −0.58u and −1.12% ROI.
The contrast is large, but it doesn't mean “Challenger is always better.” The sample mixes different tournaments, rounds, surfaces and markets. It's useful for opening a line of monitoring, not for applying a blind category filter.
By market
- Set handicap: 64 picks, −5.47u over 339u.
- Correct score: 5 picks, +6.48u over 30u.
- Game handicap: 7 picks, −6.69u over 24u.
- Total games: 2 picks, +7.87u; minimal sample.
- Total sets: 2 picks, +6.20u; minimal sample.
Both totals ended 2W–0L, but four observations between them don't justify reopening markets or raising stakes. The real volume remains in the set handicap.
Deep Signals: three systems, three profiles
July's base contains Deep Signals from day 1. Winner Elite looks for high-confidence favorites; Tapado IA looks for less obvious odds; Combi Elite daily combines the two favorites that best meet its rule. All are settled as match winner.
Winner Elite
32 signals were activated: 23 won, 7 lost and 2 voids. Across the 30 decisions, WR was 76.7%, average odds 1.30, and the result came out practically flat: −0.01u, ROI −0.03% and DD −3.84u.
The surface pattern was very uneven: on clay it produced 16W–2L, +2.93u and 16.28% ROI; on hard, 7W–4L and −1.94u. The month's close wiped out Winner Elite's provisional profit and confirms that a high WR isn't enough when working with low odds.
Tapado IA
65 signals were recorded: 33W–27L and five voids. Across the 60 decisions it adds up to +2.70u, WR 55.0%, average odds 1.94, ROI 4.50% and DD −7.91u.
Here too clay led: +7.20u and 18.46% ROI over 39 decisions. Hard closed with −3.41u and −21.31% ROI over 16, while grass contributed −1.09u. The rule became more selective during July: confidence exactly 60 and odds between 1.50 and 2.50.
Combi Elite
15 daily combos were formed and all are decided: 11W–4L, WR 73.3%, average odds 1.62, +2.50u, 16.67% ROI and DD −4u.
The opening sequence was very strong, but the close ended with four losses in the last five combos. That drop is a reminder of why a combo multiplies the odds, but also makes the entire entry depend on both legs being correct.
Combined reading
Winner Elite and Tapado IA together add up to 97 straight signals: 90 decisions, 56W–34L, seven voids, +2.69u and 2.99% ROI. Adding Combi Elite as an independent system, the tracking reaches 112 entries: 105 decisions, 67W–38L, seven voids and +5,19u.
Premium and Deep Signals can't simply be mixed together: they use different markets, prices and stakes. A −1.5 sets pick can fail even if Winner Elite gets the winner right. July specifically showed why it's worth measuring the quality of the match read and the demands of the selected market separately.
Premium + Deep Signals: the combined snapshot
If all systems are considered as one portfolio —Premium with its official stake and each Deep Signal at 1u flat— July's current cut looks like this:
The aggregate contains 77 decided Premium positions and 105 decided Deep Signals entries: 182 monetary decisions. There were also ten voids —three Premium and seven signals— and no entry remains pending.
The sum doesn't mean every user executed every entry, or that risk was concentrated at the same time. It's an accounting reconstruction of the full portfolio using a common unit.
What July leaves behind
- Premium ends positive, but not comfortably: +8.39u and 2.05% ROI with a DD of −23.38u.
- Protection decided the month's sign: +11.11u versus −2.72u from the main entry.
- V4 must be judged over time: its simulated track record is solid, but only eight real executions exist.
- Surface matters differently depending on the market: clay was negative in Premium and positive in the winner signals.
- Deep Signals close in the green, but not all of them: Tapado IA and Combi Elite are positive; Winner Elite finishes practically flat.
- The combined portfolio adds up to +13.58u: Premium and Deep Signals complemented each other, but took on 515u of cumulative exposure.
- July isn't rewritten: the improvements affect future decisions, not statistics already published.
The purpose of a monthly close isn't to prove the model is always right. It's to locate where it has an edge, where it's paying too much risk, and which hypothesis deserves to survive into the next month.
FOLLOW THE REAL TRACK RECORD
August starts from zero.
The public results will keep all of July, including its losses, while V4 and the new controls are validated with matches still unknown.
Informational content for adults over 18. Betting involves risk of loss. Past results and retrospective simulations do not guarantee future results.