It counts; it doesn't forecast. Every close, Abacus strips the market out of each name, scores what's left from 0 to 100, reads which of four tapes we're in, and lists only the trade shapes that have paid in that tape before, each with its replayed number, its tested exit, and a size. Then it marks its own book.
Named for the Bernoullis, who first wrote down that the value of a gain depends on who holds it, and that a long enough run of trials tells the truth. Bernoulli keeps everything Archimedes proved and adds only what beat it in replay: a run is listed only when the name's longer trend agrees with it; a weak name is shorted only against its own sector; the leader-vs-laggard pair: the first book's drag, is retired for a pair built on timing, the fresh riser against the stale leader; and a dislocation, where the first generation stood aside, now has shapes of its own. Judged the way every change is judged here: against buy-and-hold, on the same window, return and risk together.
The first generation, live from the launch on June 10, 2026 to the day Bernoulli took over. The name is for the man who measured a crown by the water it pushed aside, the engine's whole move: take the market out of a stock and weigh what is left. Everything it introduced is still running; Bernoulli is Archimedes with four rules added and one retired.
Not named until it has earned a rule. Three questions are open. Whether the days around a company's report deserve a shape of their own, or only a warning. Whether a theme: the AI supply chain, say, behaves as a sector of its own, with its own fund to hedge against. And whether the book should carry less in a dislocation and more when the tape is clear, rather than the same twenty slots in every weather. Each is replayed to 2008 and scored on the desk's own book after costs, against buy-and-hold on the same window; what survives ships under a new name, what fails goes on the timeline as a finding set aside.
Generations are evolutionary, never a restart: a new name means a rule was added or changed because the replay and the book said so, never a coat of paint. Each is named for someone who counted value before it was a discipline, all ages, women and men.
Every trade on the desk carries one tested number. This is how it is made, what the desk's own book has done with it, and how the engine changes, including what was tested and set aside.
Most of a stock's daily move is the market. The engine subtracts it and scores what is left, 0 to 100, as Alpha. A trade is a shape drawn from that score: a fresh strength run in a name whose longer trend agrees, a weakness run against its own sector, a cycle pair, the dislocation bounce, or in a dislocation a sector pair, with an entry, a hedge, an exit and a size already decided. The number beside it is the average outcome of that shape, in the current tape, at its exit, replayed walk-forward over 2019 → 2026 with the tape read at each entry and no knowledge of what came after. Beside it: how often it finished up, how long it was held, and the worst one-in-twenty. A shape whose number is negative in the current tape is not offered, it is here, in red, not on the desk. A shape whose number is positive but too small to clear costs is not offered either; the table says which.
| shape | expression | Trending | Flattoday | Grinding bear | Dislocation |
|---|---|---|---|---|---|
| Strength run | long · plain · with the longer trend · listed sectors | +0.82%as taken · 42% · n 1681 · offeredall that qualify +0.68% · 9d · worst -7.5% | +1.15%as taken · 45% · n 827 · offeredall that qualify +1.07% · 9d · worst -7.1% | +1.27%as taken · 43% · n 159 · offeredall that qualify +0.48% · 9d · worst -7.6% | -0.38%42% · 8d · worst -12.7% · not offered |
| Weakness run | short · vs its sector fund · against the longer trend | +0.28%as taken · 35% · n 606 · offeredall that qualify -0.05% · 9d · worst -7% | +0.15%as taken · 38% · n 756 · offeredall that qualify +0.11% · 9d · worst -6.2% | -0.46%32% · 7d · worst -6.1% · not offered | -0.64%33% · 7d · worst -8.5% · not offered |
| Cycle pair | fresh riser vs stale leader · same sector | -0.09%41% · 5d · worst -8.6% · not offered | 0.00%42% · 4d · worst -8% · not offered | +1.45%as taken · 48% · n 86 · offeredall that qualify +0.62% · 5d · worst -5.8% | +0.36%as taken · 41% · n 49 · offeredall that qualify +0.80% · 4d · worst -5.7% |
| Sector pair | a sector fund in its own trend vs one in its own slide · dollar for dollar · out when either tape turns | +0.66%as taken · 57% · n 265 · not offered | +2.28%as taken · 57% · n 377 · not offered | -4.04%as taken · 25% · n 67 · not offered | +4.53%as taken · 58% · n 52 · offered |
| Dislocation bounce | weakest tenth of a sector · vs the sector fund · 5 sessions | +0.06%56% · 4d · worst -7.1% · not offered | +0.05%55% · 4d · worst -6.8% · not offered | +0.24%58% · 4d · worst -6.6% · not offered | +1.18%as taken · 60% · n 632 · offeredall that qualify +0.62% · 4d · worst -8.3% |
a shape is offered in a tape where the trades the desk takes (the top-ranked ones, after costs) test positive; the cell beside it is the average of every trade that qualifies; the book is also replayed with and without each opening, and an opening that makes the book worse stays out whatever its cell. One more rule since 2.0.1: a strength run is not listed from a sector that is in its own dislocation, the sector fund read by the same four-state tape as the index, while the index is not.
Exits: Alpha closes under 60, or day 21 (longs) · Alpha closes over 40, or day 21 (shorts) · either leg's Alpha under 60, or day 21 (cycle pairs) · either sector's own tape leaving its state, or day 21 (sector pairs) · the five-day move vs the sector back to zero, or day 5 (the bounce). The sector pair's cells are positive outside a dislocation too, but the book as a whole tested worse with it there, so it is offered in a Dislocation only. The bounce tests positive in every tape but two stock legs cost more than its number outside a dislocation; the cycle pair is offered where it tested positive, Grind and Dislocation.
| sector · long runs | Trending | Flat | Grinding bear | Dislocation | all tapes | listed? |
|---|---|---|---|---|---|---|
| Materials XLB | +0.46%42% · 9d · worst -7.6% | +0.83%46% · 9d · worst -6.5% | not enough history | not enough history | +0.47%43% · 9d · worst -8.2% | yesruns offered |
| Energy XLE | +0.77%34% · 8d · worst -6.3% | -0.66%30% · 7d · worst -8.4% | not enough history | not enough history | +0.02%34% · 8d · worst -8.3% | noruns not offered |
| Financials XLF | +0.70%46% · 9d · worst -5.7% | +0.71%47% · 8d · worst -5.6% | not enough history | not enough history | +0.71%47% · 9d · worst -5.8% | yesruns offered |
| Industrials XLI | +0.51%39% · 9d · worst -6.4% | +0.30%41% · 9d · worst -7.6% | not enough history | not enough history | +0.53%41% · 9d · worst -6.7% | yesruns offered |
| Technology XLK | +0.99%44% · 9d · worst -9.8% | +1.97%50% · 9d · worst -8% | +0.62%44% · 9d · worst -9.3% | +0.42%46% · 9d · worst -11.9% | +1.31%47% · 9d · worst -9.3% | yesruns offered |
| Staples XLP | +0.04%38% · 9d · worst -3.6% | +0.26%40% · 9d · worst -4.2% | not enough history | not enough history | -0.05%38% · 8d · worst -4.5% | noruns not offered |
| Health care XLV | +0.42%40% · 9d · worst -4.6% | +0.84%46% · 10d · worst -5.2% | not enough history | -0.70%39% · 7d · worst -10.7% | +0.42%41% · 9d · worst -5.2% | yesruns offered |
| Consumer XLY | +0.74%40% · 9d · worst -7.4% | +0.89%44% · 8d · worst -7.5% | not enough history | not enough history | +0.80%42% · 9d · worst -7.3% | yesruns offered |
Energy and Staples runs tested flat pooled over 2019 → 2026, not enough to pay for the trade; the desk does not list them. A name in those sectors can still be opened and hedged by hand.
| shape | expression | Trending | Flat | Grinding bear | Dislocation |
|---|---|---|---|---|---|
| Strength run | long · plain | +0.62%41% · 9d · worst -7.4% | +0.63%43% · 8d · worst -7.5% | +0.88%48% · 8d · worst -8.2% | +0.52%47% · 7d · worst -11.2% |
| Strength run | long · vs the index | +0.37%36% · 9d · worst -6.4% | +0.11%35% · 8d · worst -6.5% | +0.17%37% · 8d · worst -6.6% | -0.57%33% · 7d · worst -7.7% |
| Strength run | long · vs its sector | +0.35%38% · 9d · worst -5.9% | +0.07%37% · 8d · worst -5.9% | +0.16%39% · 8d · worst -5.8% | -0.55%32% · 7d · worst -7.1% |
| Weakness run | short · vs the index · half size | +0.12%36% · 8d · worst -6.3% | +0.12%37% · 8d · worst -6.2% | -0.36%32% · 7d · worst -6.3% | -0.65%32% · 6d · worst -8.8% |
| Leader vs laggard | pair · wide gap | -0.17%40% · 16d · worst -15.8% | +0.20%43% · 16d · worst -15.1% | -0.38%37% · 16d · worst -14.8% | +0.09%46% · 16d · worst -15.7% |
| Leader vs laggard | pair · mid gap | -0.16%41% · 15d · worst -13.5% | +0.21%42% · 16d · worst -13.3% | +0.46%45% · 16d · worst -12.6% | -0.02%39% · 15d · worst -13.6% |
| Leader vs laggard | pair · narrow gap | -0.45%40% · 15d · worst -14.4% | +0.29%43% · 16d · worst -10.9% | not enough history | not enough history |
Exits: Alpha closes under 60, or day 21 (longs) · Alpha closes over 40, or day 21 (shorts) · three Narrowing days in a row, or day 21 (pairs). Pair buckets: gap: wide ≥55 · mid 42–54 · narrow <42.
Bernoulli Archimedes buy & hold (SPY) · 100 = 2019-01-02 · log scale · replay, after costs, 20 slots
| Bernoulli | Archimedes | buy & hold | |
|---|---|---|---|
| return / year | +21.7% | +7.1% | +14.6% |
| volatility | 13.0% | 9.1% | 19.4% |
| Sharpe | 1.67 | 0.78 | 0.75 |
| Sortino | 2.54 | 1.13 | 1.05 |
| worst drawdown | -11.8% | -12.2% | -41.7% |
| longest underwater | 156 sessions | 478 sessions | 512 sessions |
| months up | 72% | 58% | 66% |
| best · worst month | +20.2% · -6.4% | +8.4% · -6.0% | +12.0% · -13.9% |
| 2022 | -4.8% | -1.0% | -21.7% |
| market exposure (β) | 0.35 | 0.14 | 1.00 |
| steadiness | 0.90 | 0.89 | 0.91 |
| total | +425.0% | +72.0% | +206.0% |
Replays: each generation's rules on its own universe, 20 slots, after average trading costs (2 bps a side on a stock, 1 on a fund, included for fairness; costs vary by broker and liquidity). SPY is buy and hold. Weekly equity points, 100 = start; statistics on daily marks.
The claim is not that the book beats the index in a bull year, in 2019 and 2024 it did not. The claim is that over the whole window it earned as much or more with two-thirds of the swings, a third of the worst drawdown, a shorter wait for a new high, and less than half the market exposure. That is the standard every change to the engine is held to: return and risk together, never return at any cost.
| 2019 | 2020 | 2021 | 2022 | 2023 | 2024 | 2025 | 2026 | |
|---|---|---|---|---|---|---|---|---|
| Bernoulli | +16.4% | +56.0% | +39.1% | -4.8% | +18.1% | +9.8% | +16.0% | +15.3% |
| Archimedes | +1.5% | +18.1% | +17.7% | -1.0% | +4.3% | +5.6% | +3.0% | +5.0% |
| buy & hold | +25.3% | +15.0% | +23.9% | -21.7% | +21.7% | +20.9% | +15.1% | +11.5% |
A run is a temporary state, most end within two weeks. Runs the engine's checks agree on last about twice as long, in every tape. The clock on each trade is drawn from these curves. It is a statement about time, not about price.
checks agree all runs most agree checks split
| generation | live | sessions | book | buy-and-hold | max drawdown | closed · a trade | open |
|---|---|---|---|---|---|---|---|
| Archimedes | Sep 23, 2026 → Sep 30, 2026 | 6 | -0.6% | -1.4% | -0.8% | 29 · +1.19% | 0 |
| Bernoulli current | Oct 1, 2026 → | 32 | +5.9% | -1.7% | -3.2% | 49 · -1.46% | 20 |
Buy-and-hold is the S&P 500 over the same sessions. A position keeps the exits of the generation that posted it, so a generation's first weeks carry the previous one's positions to their own exits; nothing is re-graded when the rules change. The closed average is per trade, before slot weighting.
| month | sessions | book | buy-and-hold | difference | closed | finished up |
|---|---|---|---|---|---|---|
| November 2026 so far · through Nov 13, 2026 | 10 | +0.1% | -0.8% | +0.9% | 25 | 52% |
| October 2026 | 22 | +5.9% | -0.9% | +6.8% | 44 | 41% |
| September 2026 | 6 | -0.6% | -1.4% | +0.8% | 9 | 11% |
Book is the daily marks of the whole house book summed over the month, after costs, 20 slots; buy-and-hold is the S&P 500 over the same sessions. Closed counts trades whose exit fell in the month, at the generation that posted them. A month with few sessions says so.
Every settled day, what the rules opened and at which closes, what they closed, why, at which closes and at what Alpha, and the day's mark are written down once, hashed together with the previous day's hash, and the hash is published the same evening. The record itself unseals after 30 sessions: long enough for every position in it to have closed. A day cannot be edited afterwards without breaking every hash after it, and a position cannot be added after the fact: its day already has a hash. Chain since 2026-10-01 · 32 days · latest 0419e227999651d8… (2026-11-13)
Check it yourself: sha256(prev + "\n" + record) over the published record string, byte for byte. The chain lists every day's hash and its predecessor; the first unsealed record is a day's record in full: the string the hash was taken over. Trade ids in a record match the ids on the desk and in your tracker.
Any book can be copied partly by holding a few index funds: the market itself, small companies, cheap companies, last year's winners. This table measures how much of the replayed book those four would have copied, day by day, and what is left over. The last column is the engine's own part: what Abacus added. A book that is secretly just the market, or just a style, shows a last column near zero.
| window | with the market | small names | cheap names | recent winners | copied by the four | engine's own part, a year |
|---|---|---|---|---|---|---|
| Replay, 2019 to Sep 2026Bernoulli rules | 0.37 | +0.15 | -0.03 | +0.08 | 39% | +14.5%t-score 4.1 |
| Replay, 2019 to Sep 2026Archimedes rules | 0.14 | -0.03 | 0.00 | +0.04 | 9% | +4.7%t-score 1.5 |
| Live record since Sep 23Bernoulli | the live row appears once the record has 250 sessions (37 so far); a shorter window says more about the quarter's market than about the book | |||||
In words: over the replay the Bernoulli book moved with the market at about a third of the market's size, leaned a little toward small names and recent winners, and after all of that +14.5% a year remained that none of the four funds would have given you. That is Abacus.
The t-score (the t-statistic) is how far a result sits from zero, measured against its own day-to-day noise. For the engine it is the check that the remainder belongs to the rules and not to a good stretch: 2 is the usual bar for taking a result as real; Bernoulli's 4.1 is well past it. For a trader it is the number to ask for whenever someone quotes an alpha; without it, an alpha is a return with no weight behind it.
The four: the S&P 500 (SPY); the 2,000 smallest listed companies (IWM); large companies priced low against their earnings and book value, against those priced high (IWD against IWF); the large companies that rose most over the past year (MTUM). 2019-01-02 to 2026-09-02, dividend-adjusted; anyone with the curves above can redo the sums. Manual: Returns are not the number.
A rule must be positive over the whole window, not in a chosen slice.
Positive in a clear majority of years, and no single year outweighs the good years together.
The desk's own book, run with the change after average trading costs, must be better for it.
Return and risk together, on the same window. A change that buys return with more risk fails, however good the return.
Findings that fail are logged too, the failures are half the record.
The gate · Shorts vs the sector · The cycle pair · The dislocation bounce · Sectors that earn it · The verifiable record · Pockets (2.0.1) · Judged as taken (2.0.1) · The sector pair (2.5)
A new shape, offered in a Dislocation only: long a sector fund whose own tape is Trending against a sector fund whose own tape is a Grind or a Dislocation, dollar for dollar, two at most, out when either sector's tape leaves its state or on day 21. A dislocation is when sectors disagree most, and the exit on the tapes re-agreeing is what makes the trade pay; held to the cap it tested negative. Replayed 2019 to today after average trading costs, twenty slots: 21.7% a year at 13.0% volatility, Sharpe 1.67, worst drawdown −11.8%, market exposure 0.35, against 19.7% / 12.6% / 1.57 / −15.5% / 0.39 before. The shape's own cell as taken: +4.5% a trade over 52 trades since 2019, +3.5% over 93 since 2008.
Effect Better return and a smaller worst fall in both windows with less market, every other shape's cell unchanged; the years that changed are the three dislocations (2018, 2020, 2022), all for the better. Said plainly: 93 trades across three episodes, with energy one leg in most of them. The first rule the desk's scorecard has passed since it was written.
The replayed book's daily marks set against four public funds: the market, small names against large, cheap names against expensive, recent winners against the market. The question is whether the return is something a few index funds would have given you. Over 2019 to today the Bernoulli book moves with the market at 0.37 of its size, leans a little to small names (+0.15) and recent winners (+0.08), and the four funds copy 39% of its daily moves; what is left is +14.5% a year, 4.1 standard errors from zero. Archimedes over the same window: market 0.14, +4.7% a year left over at 1.5.
Decision No rule changes. The reading becomes a standing check: a change is taken only when Abacus's own part, the residual, holds or improves while the market loading does not rise, on the full window and in each era. The table is on this page; the live record joins it at 250 sessions.
Two rules. The listing rule now reads the number of the trades the desk takes, the top-ranked ones after costs, rather than the average of every trade that qualifies. The two differ in sign in one place: hedged shorts in a trending tape (slightly negative pooled, +0.28% as taken), so hedged shorts are offered in Trending as well as Flat. And a strength run is not listed from a sector that is in a dislocation of its own while the index is not, the sector fund read by the same four-state tape as the index.
Effect Replayed 2019 to today after average trading costs, twenty slots: 19.7% a year at 12.6% volatility, Sharpe 1.57, worst drawdown −15.5%, market exposure 0.39, against 20.3% / 12.9% / 1.58 / −15.5% / 0.42 before. A little less return for less market exposure, and better bad years (2018 −1.6 to −0.6, 2022 −9.3 to −7.7). On the long window (2008 on) Sharpe 0.80 to 0.82, worst drawdown −22.4% to −22.1%, 2018 −7.3% to −3.9%. Without the hedged shorts in trending tapes the same book replays at Sharpe 1.31 with a worst drawdown of −31.8%: they are the book's ballast, not a return source.
The second generation. A run is listed only when the name’s longer trend agrees with it; a weak name is shorted only against its own sector fund; the leader-vs-laggard pair is retired for the cycle pair (a fresh riser against a stale leader), offered in a Grind and a Dislocation; the dislocation bounce is added, the weakest tenth of a sector held against its sector fund for five sessions at most; runs are listed only in sectors whose runs tested positive; seventeen names and Materials join the universe; the sector cap comes off. Replayed 2019 to today after average trading costs (2 bps a side on a stock, 1 on a fund, included for fairness; costs vary by broker and liquidity), twenty slots: 20.3% a year at 12.9% volatility, Sharpe 1.58, worst drawdown −15.5%, market exposure 0.42, against buy-and-hold’s 14.6% at 19.4%, 0.75 and −41.7%.
Effect More return at less risk than the index over the window, and a third of its worst fall, the standard every generation is held to. The book keeps one record across generations; positions keep the exits of the generation that posted them.
The universe was widened to about 260 names (every sector thickened, three sectors added) and the Bernoulli rules replayed on it. The book got worse: more names meant more marginal runs competing for the same twenty slots. A hand-picked set of additions chosen on their own past record did no better out of sample.
Decision Set aside. Seventeen names and one sector were added where the shape has its edge; the rest were not. Selection carries part of the edge and stays a decision, not a dial.
Four other ways to fill the twenty slots were replayed: ranking by score level, by how many checks agree, by a blend; sizing by agreement; carrying more or less gross by tape. None beat ranking by the five-session rise of Alpha; the differences were inside noise.
Decision Set aside. Bernoulli ranks as Archimedes did.
Adding a standing index position to the book raises its return, and its volatility, its worst drawdown and its market exposure by more. Every version that beat the base book on return lost to it on the whole table.
Decision Set aside on the standing rule: a change that buys return with risk fails. The book stays a stock-selection engine; anyone who wants more market can hold the index beside it.
The dislocation bounce tests positive in every tape on paper, and a grinding bear looked like the next place for it. With two stock legs and costs charged, the book was better without it there: the number outside a dislocation is real but too small to pay for the trade.
Decision Set aside. The bounce is offered in a Dislocation only.
Whether the market’s recent preference for strong names over weak ones (its rotation) predicts how the next runs go was tested walk-forward. It does, a little, pooled; not enough to act on, and it would have said nothing useful in the summer of 2026 when it was most wanted.
Decision Set aside as a rule; kept as a caution the desk may show. Nothing in the score or the list changed.
The three-per-sector cap that helped the first generation’s book was replayed on the Bernoulli rules. With the gate and the sector rule already keeping the marginal runs off the list, the cap cost return without lowering the drawdown.
Decision Retired with Bernoulli. It was the right rule for the first book and the wrong one for the second, which is what a generation is for.
Alpha · The tape · Exits by state · True peers · Tested numbers · The house book
The list and the house book hold at most three long runs from one sector; the rest fall to the next shape in rank. Replayed 2019 to today on the desk's rules with the book as the objective: Sharpe 0.46 → 0.60, worst drawdown −16.9% → −13.2%, 2022 −6.7% → −2.5%, return up. A rotation is usually a sector rotation.
Effect The book stops eating a sector rotation whole. Per-trade numbers are unchanged; this is portfolio construction, not scoring.
Abacus (Alpha · Breadth · Attributed · Covariance · Utility · Score) is introduced as the engine behind every number on the desk, with Archimedes as its first generation and Bernoulli named while in study. Nothing in the rules changed; the name records what was already running.
Effect Every future change to the rules lands under a generation name on this page, so the shading tells you which rules were in force.
The desk's front door is now a short list of fully specified trades the engine has qualified (entry, hedge, exit, size), each with one tested number: the average outcome of that shape, in the current tape, at its exit, replayed over 2019 to today. Shapes whose number is negative in the current tape are not listed. The house book starts: every qualified trade, taken by the rules, after costs.
Effect The desk's rules, run as one 20-slot book over 2019–2026 after costs: about 1.0 Sharpe with runs unhedged, about 0.6 market-neutral.
Nine measures of the market's condition at entry (dispersion, rotation, leadership breadth, a leaders-minus-laggards factor, sector dispersion, the index's own return) were tested against the outcome of every leader-vs-laggard pair since 2019. None separates the years the pair paid from the years it lost. A trailing-year number predicts the next month slightly backwards.
Decision Set aside: the pair's number is an era property. It is quoted pooled, never by cycle, and the leader-vs-laggard pair is not offered in an up-tape.
The same fresh-leader entry, hedged four ways: with nothing, with the index, with its sector, with a laggard stock. Hedging with the index or the sector keeps the run's number positive in every tape but a dislocation; hedging with a laggard turns it negative in an up-tape, because laggards get bought in rotations.
Decision Runs are quoted plain with the index or sector hedge as a switch; weak names are offered only against the index; the laggard pair returns in flat and grinding tapes, where selling the weak name pays.
Shorter caps, two Narrowing days instead of three, hard stops on the spread, trailing give-backs, profit takes, and an exit on the legs' own Alpha were all replayed on the same entries. Every one lands within a fifth of a percent of the shipped rule. Exits change the shape of the outcome (the leg exit halves the worst case at a third of the holding time), not the mean.
Decision Exits unchanged. The leg exit is a candidate for grinding tapes once a third such tape has been seen.
Runs the checks agree on last about twice as long as runs they split on, in every tape. Their return is better in up and flat tapes and worse in the one grinding year on record.
Decision Agreement conditions the run clock on the trade detail. It does not move the tested number, the rank, or the size.
A stretched bullish read in a grinding or dislocated tape has its conviction trimmed and the ticket says why. Dislocation now shows whether the index is still falling or rebounding off its low, because the two halves tested differently.
Effect Removes the one cohort that lost in both 2020 and 2022 without touching entries elsewhere.
Four market states read from the index alone (Trending, Flat, Grinding bear, Dislocation), settled once a day, with a posture line for each. Twenty years of replay chose the four and their persistence rules.
Effect Every tested number the desk quotes is now conditioned on the tape at entry.
A pair that wobbles a point past its threshold stays on the board inside a grace band instead of churning off and on. Retained pairs went on to outperform fresh pair-days in testing.
Effect Board churn fell from most of the board each day to a handful; exits became real events.
Pairs leave on three Narrowing days in a row; directional runs leave when Alpha closes back through 60. Both replaced fixed holding periods after a study of every exit rule the desk could express.
Effect Kept most of the payoff in less than half the time with about half the worst case. Still the exits in use.
Peers are found by how names move together after the market's influence is stripped out, not by sector label. Some cross sector lines.
Effect Tighter hedges and, later, the finding that a tighter hedge is not a stronger thesis.
Relative-strength scoring across stocks, sectors and the market: every name's move with the market taken out, scored 0–100 as Alpha, with paired long/short ideas built from the gaps. The rules Archimedes still runs on.
Effect The first generation begins; everything above was added to it, nothing replaced.
Open the deskOperator's ManualThe WorkshopChangelogDaily Pulse
Analytical tools and historical measurements for educational and informational purposes only; not investment advice or a recommendation. Replayed figures are hypothetical: today's rules applied to past prices, after modelled costs, and no account was traded on them. Hypothetical results have inherent limitations and do not represent actual trading. Past performance, including backtested results, does not guarantee future outcomes. Trading stocks involves substantial risk of loss. About · Support · Terms · Privacy · Refunds