Why we passed on Marlins–Diamondbacks

We’re looking for an edge. So when a game shows a positive estimate, why wouldn’t we take it?

Because the estimate is only part of the decision. We also need the setup to meet its confirmation requirements, and the available price still has to qualify.

Here’s a real example from our saved records for Miami at Arizona on September 15, 2026. At one check, the estimated expected value was positive. Confirmation was still pending. At the next saved check, the price no longer qualified.

What we actually recorded

These times are Eastern Daylight Time on September 15. The source file uses UTC, which places the last six checks on September 16.

At 8:52:19 p.m., the estimate was +2.32%. The recorded status was WATCH, with the reason “price qualified, waiting for confirmation.” That was a setup to keep watching.

At 8:57:52 p.m., roughly five and a half minutes later, the estimate was −1.55%. The recorded reason had changed to “price below the required level.” At 9:12:52 p.m., it was still below that level.

All seven saved observations show WATCH, and none contains a confirmed paper-play identifier. When we say we passed here, we mean the monitored setup stayed unconfirmed in these records. WATCH is the actual system label.

A positive number isn’t the whole answer

Expected value, or EV, is an estimate based on a probability assessment and a price. It is not the chance of winning, and +2.32% does not promise a 2.32% return on this game.

The question we care about is whether the setup meets the rules at a price we can still use. An earlier positive estimate doesn’t give us a reason to ignore the next check.

That’s why this example matters. We had a moment worth watching, followed by a saved observation that no longer met the price requirement. We didn’t have a confirmed play in the records shown here.

What this example tells us

This is a look at one historical decision sequence. The seven rows are repeated checks of one game, not seven independent opportunities. They don’t capture every price change between checks.

The audit extract does not include the selected side or exact sportsbook odds. We’re leaving those out rather than reconstructing them. It also doesn’t establish whether a different setup qualified elsewhere or later.

We’re not using the final score to justify the decision. This is not a claim that passing saved a loss, or that these estimates prove a profitable strategy. It shows why a promising number didn’t become a confirmed play in this sequence.

So what happened in the game?

Miami won 4–2 in 11 innings. MLB’s game recap records the result.

We also found a separate paper experiment for this game: Miami at +130, captured at 8:07:14 p.m. EDT. That record was graded a win, with a paper profit of 1.3 units. The experiment was called “Priced control: no edge floor.” It did not require a minimum estimated edge.

That matters. The experimental bet won, while the monitored sequence shown above stayed unconfirmed. These are different research records. We can’t turn the experimental result into a claim that our monitored setup was a confirmed winner, or that passing avoided a loss.

Would a losing bet make this a cleaner story? Sure. But choosing only losing examples would tell you less about how we work. A bet can win without meeting our rules. The useful question is whether we followed those rules with the information available at the time.

Read the separate experimental record →

What you should expect from us

If we’re asking you to trust the process, we should be able to explain the quiet days too. Sometimes a price isn’t good enough. Sometimes confirmation is missing. Those reasons should be visible.

Our job is to keep the record clear: what we saw, what the rules required, and what happened next. This time, that meant continuing to watch instead of treating an unconfirmed estimate as a play.

Research only. Historical estimates and paper results do not guarantee future returns.