Same record.
Clear definitions.
The daily record and monthly table use a connected account feed when available. The comparison chart, research log, and downloads remain a dated September research snapshot. Each view identifies its observation period.
Internal validation
Quants.Space ↗ is the internal validation reference. The strategy-specific record has not yet been linked; the homepage link is not evidence of completed validation.
Connected account observations
The public feed contains normalized daily equity and aggregate trade statistics. It excludes credentials, raw account balances, and open positions. The backend combines recorded history through 31 May 2026 with account equity reconstructed from actual fills, position sizes, fees, funding, wallet balances, and mark prices from 1 June. This differs from the full-size signal model used in the archived September snapshot.
Daily metrics use completed UTC daily observations; an unfinished day is excluded. The page shows the feed’s observation date and generation time, marks stale data, and identifies when it falls back to the saved snapshot. Equity growth can include external deposits or withdrawals: Sharpe, APY, and Calmar are withheld in the connected view unless the feed verifies that the entire evaluated period has no external cash flows. Drawdown describes account equity, not a cash-flow-adjusted investment return.
When the account feed is available, the monthly table uses its daily equity rather than substituting June and July backtests. The archived research log keeps its separately labelled backtest series. The comparison chart retains its common September cutoff so all benchmark and strategy dates match. The downloads below are dated source snapshots, not exports of the current connected feed.
Archived daily snapshot · September 2026
- 10 Jul 2023–31 May 2026: 1,057 recorded account observations. Every overlapping row exactly matches the previously supplied May daily file.
- 1 Jun–11 Sep 2026: full-size reconstruction from actual trading signals, with 0.05% entry and exit fees plus funding. Signal sizes are discarded in the model. Stalker/Bybit signals are used from 22 Jul. These are modelled equity observations, not actual full-size account balances.
- July 0.0%; August +11.1423%: taken from the newer daily file. All 102 reconstructed rows from 1 Jun–10 Sep were independently reproduced from cached fills, marks, and funding within 2.014 × 10⁻¹⁰ relative error.
- Latest supplied observation: 11 Sep 2026. The newer CSV/PDF contains that row, but older cached signal snapshots do not independently reproduce the final appended day. It is retained as source-reported. No strategy daily observations were supplied for 12–14 Sep.
“One running account · Jul 2023–present”, Binance API access on request, and “Vega active since Jul 2026” describe the account and version history supplied by the team. They do not extend the observed daily series beyond its cutoff or turn reconstructed returns into verified balances. The daily chart is drawn as one continuous line; its source composition remains documented here.
Connected account metrics: position duration uses actual fill timestamps for the trade sample shown with the metric. The numerical examples below belong to the archived September signal snapshot, not the connected account feed.
Archived Metric definitions
- Equity growth: final daily equity ÷ initial equity − 1. The record is indexed to 1.0 on 10 Jul 2023. Equity changes are not independently established cash-flow-adjusted investment returns.
- Average drawdown: mean of every daily equity ÷ running daily peak − 1, including days at the peak. This is a time-weighted average of daily drawdowns, not the average depth of drawdown episodes.
- Maximum drawdown: minimum daily equity ÷ running daily peak − 1. Daily closes do not measure intraday losses.
- Sharpe ratio: mean daily arithmetic return ÷ population standard deviation of daily returns × √365. Risk-free rate is zero. Flat days are included.
- Trailing 12-month Sharpe: the same calculation using 365 daily returns from 366 closes, 11 Sep 2025–11 Sep 2026. It is unavailable if a full year of observations is absent.
- Calmar ratio: full-period annualized growth (APY) divided by the absolute maximum daily drawdown. It is unavailable when maximum drawdown is zero.
- APY: (final equity ÷ baseline equity)365 / elapsed calendar days − 1. This is realized annualized growth, equivalent to CAGR for this series, not a promised yield.
- Average position duration: mean entry-to-exit elapsed time from actual fill timestamps for 12 closed Vega signal positions, 2 Aug–11 Sep 2026: 1.87745 days. Open positions are excluded. Daily balance CSVs have no entry/exit timestamps; a duration cannot be inferred from daily equity changes. This is the available fill sample, not a duration for the full historical account.
All equity metrics describe the combined daily report: recorded history followed by reconstructed signal performance. Position duration describes the stated recent Vega sample only.
Monthly calculations and archived backtest substitutions
Monthly change uses the last daily equity in the month divided by the previous calendar month’s closing equity, minus one. Where the prior close is unavailable, it uses the first available observation and marks the month partial. Zero-change months are retained.
Annual return compounds the available monthly changes. APY annualizes those results using the sum of elapsed days in each selected month’s baseline-to-close interval. The 2026 table covers 250 daily return intervals; July 28–31 are excluded because the selected July backtest ends on the 27th. Full leap years therefore have a slightly different 365-day APY than their 366-day annual return. Partial months and years show an asterisk; missing months remain blank. The 2026 row also includes the June and July backtest replacements requested for this review; it is a combined research result, not the return of one observed account.
June and July 2026: use daily_net_of_fees_CURRENT.csv from the July research folder, with backtest cells marked BT in muted purple. June +0.8666%; July +8.6241% through 27 Jul only. July 28–31 have no backtest observations and are not fabricated. August and September remain the latest supplied daily-file results from actual signals, reconstructed at full size.
Month inspection: maximum drawdown resets the running peak at the selected source’s prior-month close. Peak gain is the highest daily equity relative to that same baseline, floored at zero. The backtest books P&L at trade exits and stays flat between exits; its drawdown and peak gain exclude equity movements while trades are open. A zero event-day return can therefore mean no exit was booked that day, not absence of intraday risk. Closed trades count distinct completed positions by UTC exit month: seven in August, five through 11 September. Historical account and backtest trade ledgers were not supplied, so their counts are unavailable. These are position counts, not order counts, trading days, or notional turnover.
Archived comparison · four comparable series
- BTC hold · USD: Coinbase BTC/USD daily close divided by its 10 Jul 2023 close.
- ETH hold · USD: Coinbase ETH/USD daily close divided by its 10 Jul 2023 close.
- Strategy · ETH: supplied ETH-denominated daily equity divided by its first observation.
- Strategy · USD: normalized ETH strategy equity multiplied by normalized ETH/USD spot price for that date. This is a currency conversion of the supplied record, not a separate USD account.
The benchmark source has 1,162 consecutive UTC daily closes, 10 Jul 2023–13 Sep 2026, with no missing-day interpolation. The displayed comparison ends 11 Sep to match the strategy. Source daily dates are aligned with UTC market closes; exact account close timestamps are not established for the early report.
Price benchmarks use filled areas and their boundary lines. Strategy curves use lines. From the team’s live-launch date, 22 Jul 2026, strategy lines are solid; earlier reconstructed signal segments are dashed. This visual cutoff does not change the underlying data provenance. Each visible series shares the same start date, normalization, and chart scale. Every series starts at exactly 1.0. Axis ticks are drawn at their labelled numeric values, and a 1.0 reference makes the common baseline explicit in both linear and log views. Coinbase Exchange candle documentation ↗
Major-move selection
Compute daily BTC and ETH close-to-close changes. Keep days where either absolute change is at least 8%; rank by the larger absolute move; select the 12 largest while excluding dates within seven days of a larger selected move. The price-only 19 Aug 2026 rebound has been removed as requested, leaving 11 chronologically displayed events. Selection is mechanical, not a claim to identify every important news event.
The sidebar evaluates every event using the close dated immediately before the event through the close dated three days after it: four daily returns in total. BTC, ETH, and strategy use identical start and end dates, and all gains and losses are compounded. Strategy is shown in both USD and ETH, with the total crypto market-cap change shown separately for the event day. Complete coverage is required; missing endpoints do not silently shorten a window. Strategy USD window change = (1 + strategy ETH window change) × (1 + ETH/USD window change) − 1. The market series uses UTC closes; historical strategy dates have no verified valuation timestamps, so date alignment is a convention, not proof of simultaneous marks. Every context sentence links to a primary source; retrospective timing is not proof that the cited event caused the full move. Research equity uses the daily report before 1 Jan 2026, chains actual backtest daily returns from 1 Jan through 21 Jul 2026, then chains reported live-signal returns from 22 Jul onward. This fills the early-2026 flat period without inventing daily observations. The backtest segment is purple and dashed through 21 Jul. The 21→22 Jul return interval uses the daily report and a solid line. July’s separately labelled monthly backtest sample still ends 27 Jul; that sample is distinct from the research timeline’s live cutoff. The resulting level is a combined research index, not account equity; event strategy returns use every daily observation in the stated window. Wider windows can capture later booked P&L but also unrelated position changes and valuation-timing effects. Exact entry/exit attribution is not established: available historical fills cannot be reliably joined to the reported account, and the February 2026 backtest trade ledger is absent. The model records no change around 5 February until 24 February; that later booking cannot be attributed to the crash without its trade records.
Event-window check
All 11 events were checked at event day, +1, +3, and +7 days using the same preceding-day baseline. Through +1 or +3 days, ETH-denominated strategy returns improve for seven events and remain unchanged for four. Through +7 days, seven improve, three remain unchanged, and one worsens. These comparisons show window sensitivity; they do not establish that the event caused a trade profit.
The review uses one fixed +3-day endpoint for every event. It is an exploratory view, not a claim that three days is optimal or that every position has closed. USD returns can move differently from ETH-denominated returns because ETH/USD changes during the window.
| Event date | Strategy USD · day | BTC USD · +3 | ETH USD · +3 | Strategy USD · +3 |
|---|---|---|---|---|
| 2023-11-09 | +14.45% | +4.06% | +8.32% | +16.79% |
| 2024-03-20 | −0.71% | +3.38% | +5.48% | +3.52% |
| 2024-05-20 | +19.38% | +2.53% | +23.12% | +39.41% |
| 2024-08-08 | +14.50% | +6.48% | +9.04% | +9.04% |
| 2024-11-06 | +20.68% | +10.57% | +29.14% | +58.63% |
| 2025-03-03 | −14.73% | −4.61% | −12.55% | +4.11% |
| 2025-04-09 | +13.41% | +11.83% | +11.68% | +11.68% |
| 2025-05-08 | +14.19% | +7.29% | +38.84% | +42.90% |
| 2025-08-22 | +14.36% | −2.09% | +3.61% | +3.61% |
| 2025-10-10 | −15.49% | −5.29% | −2.89% | +16.01% |
| 2026-02-05 | −14.95% | −3.72% | −2.57% | −2.57% |
All 44 window calculations were independently recalculated from normalized daily equity and Coinbase prices. There are no missing dates or source transitions within these windows. Historical cash-flow adjustments and exact strategy valuation times remain unverified.
Event-day crypto market value
The event-day row uses total cryptocurrency market capitalization in USD from the CoinMarketCap public global chart, including stablecoins. It is the full tracked market, not a BTC/ETH average or an index return.
For an event dated D, change = market cap at D+1 00:00 UTC ÷ market cap at D 00:00 UTC − 1. These are exact daily midnight snapshots spanning the event’s UTC calendar day; no interpolation is used. All 11 events have both endpoints. This one-day market-cap change is separate from the four-day BTC/ETH and strategy comparison above.
The chart’s public historical data was retrieved on 14 Sep 2026. Four recent samples matched the current daily chart within $1 rounding. Supply and listing changes can affect total capitalization as well as asset prices.
Testing scope
The team reports strategy testing on OKX, Bybit, Hyperliquid, and Binance, with test scale of 1,205,604.00 USDT and 216.687 ETH. These supplied testing figures are neither the performance denominators nor representations of current account balances.
Download provenance
Dailies: dailies (sep11).csv · 1,160 daily rows · 10 Jul 2023–11 Sep 2026.
Report: ErgoCap_Vega_v4_2026-09-12.pdf.
Daily-file SHA-256: 8d883bca4595a0756b17f07595537615a27c8f8f0f34435ac980f0a858363d36
Backtest: daily_net_of_fees_CURRENT.csv · 783 consecutive daily rows · 5 Jun 2024–27 Jul 2026. SHA-256: 8d82243d33a5550f3f34a521fd171a238d2a36111cd20e4075b5eec5a723573a. Referenced trade-level backtest ledgers were not found in the supplied folder.
The downloads are the supplied source documents. The charts contain normalized values; no API credentials are included.