Prediction Track Record

• Graded as published: a call is scored as it went out. The call itself is never edited, withdrawn, or removed once published.
• Grading bugs are fixed rather than left in place, and every fix is listed under Corrections below, with its date and what it did to the total.
• Entry and settlement: each play settles on its own entry session.
• Voided calls: when a company reported earnings outside its holding window but within 21 days of it, the call is voided and counts as neither a win nor a loss. Price drift alone cannot settle a call whose catalyst never arrived. A report a whole quarter away says nothing about that window, so those calls settle normally.
• What else is absent, and why: a call is not counted while it waits on its expiry. Past expiry it is scored, unless something needed to settle it is missing, in which case it stays out rather than being guessed at. That can be the published strike, the closing price, or, for a call on a company that is not itself reporting that week, which company’s report it is judged against. The weekly analyses also record names we looked at and passed on, and a decision not to trade is not scored as a call.
• P&L basis: spread P&L is modeled from real closes around expiration. Win or Loss is the sign of that modeled P&L.
• Stake per call: each spread is modeled as risking its conviction tier’s budget as its maximum loss, and a spread with no tier risks the Tier 3 budget. Single-leg long calls and puts carry no dollar figure, so they have no stake. From 7 September 2026, halo trades are the exception. A halo is a momentum trade on a company that is not reporting that week, and it risks the maximum set in its note’s sizing table, or $400 when the table sets none. Halo trades entered before that date are staked like any other spread.
• When a call is scored: at its expiry, always. Some notes instruct an earlier exit, usually a one-event trade told to close the day after the catalyst reports. That instruction does not move the date this page scores against, because the exit price is not something we published in advance and could not be checked afterwards. So a call can already be settled in that week’s written review, which books it at the exit it was told to take, and still sit open here until its expiry passes. Neither number is wrong. They are answering different questions, and this page only ever answers the one it can show a closing price for.
• Where the modeled credit is weakest: where a note states the credit it expected, the spread is scored on that figure. Where it states none, and most do not, the credit is a fixed fraction of its width, calibrated in July 2026 from credits quoted in our own notes at the strikes those notes placed. When a strike is re-derived to a different distance from the price, that assumption no longer fits it, and a strike moved further out of the money is credited more than it would really have earned. QCOM in the week of 28 July is the clearest case: sound as a win, because the option expired worthless, but its modeled dollar figure is generous.
• Losses included: every settled call appears, with sample sizes shown.
• Scope: this ledger counts published SeerForecast calls. The in-app Model Learning panel counts every lane the model trains on, including the retired OpenAI engine's frozen history, so its totals are larger and its win rate differs from this page's.
• Benchmark: the equity curve overlays SPY over the same period.
• Audit it yourself: the full ledger is downloadable as CSV.
• Paper indices: three portfolios we build each week by fixed rules from our published stock ratings, not from these earnings calls. Their weekly returns against SPY are at /indices.
• Archived off-site: we file copies of this page with the Internet Archive (opens in a new tab) after each week's settlement, and the calendar there shows every capture. We cannot alter a stored copy, and we will not ask for one to be removed, so the promise above that a published call is never edited is one you can check against a record we do not host. What moves between two captures is the grading, never the call, and every correction is listed below.

Corrections

On 1 August 2026 we fixed three bugs in how the ledger read our own published notes. None of them changed a call. All three changed how a call was scored. Together they moved the published total from $12,277 to $12,460, a rise of $184, and the settled record from 80 wins and 27 losses to 79 and 27.

• Strike prices were being read out of dates. An entry line reading "Mon 7/6 to Tue 7/7" was parsed as a $7 strike, and a $62 stock cannot close below a $7 barrier, so the play scored a win it could not have lost. Three calls were priced this way. Regrading them against their real strikes lowered the total by $939, and turned one published win into a loss.
• One call carried another play's structure. A parser fault copied a neighbouring alternate onto a halo trade, which was then scored as a spread it never was. Read correctly it is a long call, and long calls carry no modeled P&L, so it now shows an outcome with no dollar figure. This lowered the total by $188.
• Strikes were read from the stale half of the note. Each note states placement twice, as a percentage below or above spot and as an approximate dollar strike worked out on the Friday before. The dollar figure goes stale when the stock moves before entry. On QCOM the reference was 7% high by Monday, which left a strike the note placed well clear of the price sitting on top of it. Grading against the stated percentage instead raised the total by $1,311, and turned one published loss into a win.

The last of those three works in our favour, which is why it is written out here in the same detail as the two that did not.

On 18 August 2026 we fixed a fourth, which moved no total on this page but did move what three rows show. Three calls from the week of 25 May were dated to a session the market never opened. A call on a company reporting before the open is entered the session before the print, and the code that worked out that session skipped weekends but not holidays, so all three landed on Memorial Day. The price lookup then resolved forward to the next available close. For a before-open reporter that is the session which already contains the earnings move, and the strikes are worked out as a percentage of the entry price, so they moved with it. AZO shows the size of it: entered at $3,100.11, the close after a 9% fall on its own print, when the last real session before it closed at $3,406.50.

Regrading the three against the correct session changed no outcome and no dollar figure, so it left the settled record and the published total exactly where they were. All three were wins before and after. What changed is the entry date and the strikes shown against those rows. Each of the three closed far beyond its winning strike in both placements, and a call that finishes clear of its strike books the same modeled result wherever inside that zone the strike sat, so moving the strikes moved no dollars.

On 25 August 2026 we fixed a fifth. The ledger could not see halo plays at all. A halo is a momentum trade on a company that is not reporting that week, which is the whole point of the structure: its options carry no earnings premium. The code that reads our published notes only recognised a bolded ticker heading or a wide table row, and a halo is written as neither, so a halo whose peer did not also happen to report that week was absent from this page while the week’s own written review scored it.

Reading them back added four calls and moved the published record from 90 wins and 33 losses to 91 and 33, over 124 calls instead of 123. One of the four is settled and it is a win. One is still waiting on its expiry. Two are void.

The first void is the case worth stating plainly. That call was published and never entered: the live debit exceeded its own execution gate before entry, and the week’s review records it three separate times as not traded and excluded from the tally. Read back naively it would have settled as a win, so scoring it would have credited us with a trade nobody placed.

The second is void because its note never published an expiry. It said “one to two week” and named no date, and the ledger had settled it at four days, where it showed a win. A call graded on a date we never published is not a result, whichever way the number falls, so it is excluded and says so.

One further change sits behind those numbers. A halo is now judged on whether its anchor reported inside the holding window, not whether the halo itself did. The peer never reports that week by design, so measuring it against its own next print was removing halo calls for a reason that had nothing to do with the trade. That change applies only to the calls added here; no call already on this page was re-graded.

The one added call is a win, which is why it is written out here in the same detail as the fixes that cost us. No existing row changed its outcome, and nothing was removed.

On 28 August 2026 we made a sixth change, to how the protective leg of a spread is read. It moved no total on this page, and it moved the strikes on ten rows. A spread has two legs, and each note places them as a distance from the price it was written at. The near leg was always measured from that price. The protective leg was measured from the near leg instead (“buy the put ~4% lower”), a bare percentage that the code reconstructing these strikes does not read, so that leg fell to a default width of ours instead of the distance the note itself gave. Notes now measure both legs from the price, and the ten older rows already written that way are read as written.

On all but one of them the figure that moves is the protective leg, the one further from the price. The exception is a debit spread, where the leg that moves is the one shown first. No outcome and no dollar figure moved either way: all ten closed outside both of their strikes, and a call that finishes clear of its strike books the same modeled result wherever inside that zone the other leg sat. These strikes are still not fills, as the note under the table explains. What changed is where the protective leg's distance comes from: the published note, rather than our default.

On 14 September 2026 we made a seventh change. This one removed a row from the page.

The analysis for the week of 7 September chose Arista Networks as its halo trade. It named Vertiv as a fallback in a sentence that ended “Take one peer, not both,” and its sizing table funded Arista alone. The code that reads our notes took that sentence for a second call, because it found the ticker next to the words “call spread”. It booked a Vertiv bull call spread with no tier and default strikes, then scored it as a $2,000 loss.

The analysis recommended no Vertiv trade that week, so the row had no call behind it. We removed it, along with its card on that week’s page. Arista’s call is unaffected.

The copy of this page filed with the Internet Archive on 12 September shows the row with its loss, a total of $12,992 and a record of 104 wins and 37 losses. Without the row, that copy would have read $14,992 and 104 wins and 36 losses: $2,000 more and one loss fewer.

That week’s written review went out before the fix. Its line for the public ledger and its running tally still include the row.

This change works in our favour. That is why it gets the same detail as the corrections that cost us.

On 22 September 2026 we made an eighth change, to how a credit spread is scored. It changed no call and no outcome. It lowered the published total from $15,705 to $11,875, a fall of $3,830, and left the settled record where it was.

A credit spread collects a credit when it is opened, and that credit decides how much of the spread’s width is profit and how much is risk. Where a note stated the credit it expected, this page was ignoring it and scoring the call on a flat assumption of 32% of the width instead. The clearest case is Kroger in the week of 7 September: the note published a credit of $0.16, which is 8% of the width this page records for that spread, and the page scored it as though it had collected four times that.

Seventeen of the 140 settled calls in the portfolio curve above stated a credit we can read, and all seventeen are now scored on what they stated. Fourteen of them move. The other three finished at a full loss, and a full loss cannot move either way, because it is the whole budget whatever the credit was. One row moves up: the Analog Devices call of 22 May stated $4 on a $10 width, a larger credit than the assumption, so reading it raises that row.

This does not make the total a measured number. Most notes state no credit at all. On the day of this change 91 of the 108 settled credit spreads still carried the flat assumption, and the credits our own notes quoted through August and September ran well below it. So the figure above is still a modeled one, and still higher than a fully measured figure would be. Re-fitting that assumption is a separate change, and it will get its own entry here.

On 4 October 2026 we made a ninth change. Like the seventh, it removed a row from the page.

The analysis for the week of 28 September passed on Accenture. It listed the company in its table of names reporting that week with no play, noting that a put spread there “models 16% of width, under the 20% credit floor”. The code that reads our notes found the ticker next to the words “put spread” and booked a bearish put spread with no tier. Accenture rose 16% on its report, so the row settled as a $2,000 loss.

The analysis recommended no Accenture trade that week in either direction, so the row had no call behind it. We removed it, along with its card on that week’s page, and the code no longer reads a play from that table at all. No other call that week changed.

The copy of this page filed with the Internet Archive on 3 October shows the row with its loss, a total of $10,206 and a record of 110 wins and 40 losses over 150 settled calls. Without the row, that copy would have read $12,206 and 110 wins and 39 losses over 149: $2,000 more and one loss fewer.

That week’s written review went out before the fix. Its line for the public ledger and its running tally still include the row.

This change works in our favour. That is why it gets the same detail as the corrections that cost us.