For Bitget rToken holders · Alpha Factory

Sell before the dividend? Usually, don't.

When a stock pays a dividend, its rToken drops overnight by about that much. So selling the night before and buying back in the morning sounds like free money. I tested it on 127 dividends. It isn't. Exnight tells you, event by event, whether it's worth it, and every call is locked in before the market moves.

Open the app → Try the calculator

How it works

It comes down to three numbers.

All three come from Bitget data. If selling first doesn't clearly come out ahead, the answer is hold.

01 · THE DROP

The price really does drop

Across 127 past dividends, by 04:00 ET the rToken had dropped by almost the whole dividend.

0.97× the dividend
02 · THE TAX

But you give up the dividend

If you sell, you don't get paid. After up to 30% US tax, that's 70% of the dividend you're walking away from.

70% kept after 30% tax
03 · THE COST

And getting out and back in isn't free

Two fees plus slippage. Whatever's left after the dividend you gave up has to cover this.

~45 bps round trip

Interactive

Try it yourself.

Drag the sliders. It's the same math Exnight runs on every event, just for one trade. Try the rBZ preset to see what it takes for selling first to pay.

Most rToken dividends are 0.1%–0.8% of the price.
The average is 0.97, but it bounces around. Exnight plays it safe with 0.60 so it doesn't trade on a hunch.
The more tax gets taken, the less you lose by skipping the dividend.
Two 10 bps fees plus 25 bps of slippage is 45 bps.

Exnight says

HOLD

Drop avoided
Dividend given up
Trading cost
Price drop you avoid by selling
− dividend you would have kept
− trading cost
Selling first vs holding

On $1,000, that's . Numbers are in basis points (1 bp = 0.01%) of the token price.

Proof

Tested on data it hadn't seen, then live.

I fitted the rule on older events and tested it on newer ones it had never seen. Then I ran it live on 16 events: each decision went into Git before the 20:00 ET cutoff and was graded after the ex-date from a minute-by-minute recording. No edits after the fact.

−32.5bps lost per event if you always sell first
13%of the time selling first actually won
39unseen events. Exnight said hold on all of them
—live events graded

What I found

  • The overnight drop is real. On average it's 0.97× the dividend.
  • On unseen data, Exnight came out 32.5 bps per event ahead of selling first.
  • Selling first lost at every cost I tried (10 to 100 bps) and every tax rate (0% to 30%).
  • Buying just to catch the dividend lost money too, in every setup.
  • Bitget's Playbook backtest doesn't credit dividends, so it makes selling first look better than it is.
  • A real $10 order filled at the quoted price, even with an empty public order book.
  • Every call is in Git before the cutoff, so nobody can change it once the result is in.

Quick glossary

Four words you'll see in the app

EX-DATE
The day the price drops by the dividend. You must hold the token before it to get paid.
HOLD
Keep the token through the ex-date. Selling first would cost more than it saves.
EXIT
Sell before the cutoff and buy back after. Only when that clearly wins after tax and costs.
NOT ENOUGH DATA
Not enough to go on, like a trade too big for the quoted price. It won't guess, so you keep holding.

Check it yourself

Look up a token, see every live call, grab the raw data.

Open the app →