The answer
Loading the result…
By 04:00 ET on the ex-date, an rToken has usually fallen by about the whole dividend. Exnight checks whether selling first is worth it after the tax you'd pay on the dividend, fees and slippage.
Read this page in 30 seconds
- 01Live test: decisions locked before each cutoff, graded after. Latest results below.
- 02Backtest: always selling first loses money. Exnight didn't.
- 03Check a token to see its decision at $1k, $5k and $25k.
01 · Live test results
Locked before the cutoff, graded after the ex-date
| Token | Ex-date | Locked decision | What happened | Selling first, per share | Recording |
|---|
02 · Backtest on unseen data
Exnight vs always selling first vs holding
| Approach | Trades | Vs holding, per event | Beat holding | Total return | Sharpe |
|---|
02 / LIVE TEST↗
Every scheduled event
High-dividend ex-dates, the decision locked for each, and the graded outcome.
03 / CHECK A TOKEN↗What should I do with my token?
Type a symbol, get HOLD, EXIT or NOT ENOUGH DATA for each trade size, with the reason.
04 / BACKTEST↗How it was tested
Fitted on the past only, tested on later events, across every cost and tax level.
05 / RECORDER↗Minute-by-minute prices
The raw price recording the grades are computed from, with its health checks.
06 / EVIDENCE↗Files and limits
Download every score and manifest. Read exactly what isn't proven yet.
↺ / CALCULATOR↗Try the arithmetic
Move the dividend, tax and cost sliders to see when selling first could pay.
02 · Live test
Every high-dividend event
Each decision is saved and committed to Git before the 20:00 ET sell cutoff, so it can't be changed later. After the ex-date, a scorer grades it from the minute-by-minute recording.
Real position
rTOWN dividend
Live test
Decisions and outcomes
| Token | Ex-date | Dividend | Locked decision | Why | Outcome |
|---|
03 · Check a token
What should I do with my token?
Look up an rToken to see its decision at each trade size, or browse every saved decision below. Click a row for the full arithmetic.
Token lookup
Enter a symbol
Any form works: rAPH, RAPHUSDT or APH.
The decision for each trade size appears here.
All saved decisions
Decisions by ex-date
FIRST VERSION
| Event | Trade size | Decision | Gain from selling first / cost, per share | Price quotes used |
|---|---|---|---|---|
| Loading… | ||||
04 · Backtest
How it was tested
The rule was fitted using only past events, frozen, then tested on later events it had never seen. Only dividends published before each decision were used.
Scored on the Alpha Factory metrics
Exnight minus always selling first
Each night's return is Exnight's return minus the return of stepping out of every event, on the same events, prices, costs and capital.
| Metric | Training period | Unseen test period |
|---|
Full record, daily returns and per-event rows: reports/backtest.md · rerun with python scripts/run_backtest.py
Unseen-data performance
Exnight, holding, and the difference
| Series | Total return | Sharpe | Max drawdown |
|---|
Assumptions
What the result means
Test periods
Training periods and the later test periods
| Test month | Training end | Test period | Price read at | Drop estimate | Test events | Sell trades | Test Sharpe |
|---|
05 · Recorder
Minute-by-minute price recording
The recorder saves each token's price every minute through the event. Scores are computed only from this saved record, never from newer data.
First recorded test · —
Loading…
Loading the saved recording and its checks.
Checking recorderLoading
Price
Recorded price and order-book coverage
Loading…
last traded pricepublic order book presentlast 240 samples
Health
Recording checks per token
06 · Evidence
Files, sources and limits
Download the scored results and manifests behind every number, and see exactly what is not yet proven.
First live recording
22 September result (first version)
ResearchEvent calendar, price study, costs and saved decisions
RecordingChecking
GradingChecking
Sources
Where the numbers come from
22 September tokens
What happened for each token
| Token | $1k decision | Dividend amount | Price before → after | Result |
|---|
Honest limits