Ending the night flat
Spread income is the salary. Inventory is the position you accidentally acquired while earning it. Manage the second or lose the first.
Why inventory accumulates
Quote both sides long enough and fills will not balance: the market drifts, one side gets hit more, and you wake up long 400 tokenized TSLA you never wanted. Your PnL is now dominated by TSLA's next move, not by your spread — you have become a directional trader by accident, at exactly the hours when direction is hardest to price.
Controls the kit enforces
| Control | Default | Effect |
|---|---|---|
max_inventory | 5 × quote size | Hard cap per pair. At the cap the kit stops quoting the accumulating side entirely. |
skew_per_unit | 0.2 half-spreads | Every unit of inventory shades both quotes toward flat, so unwinding pays better than adding. |
flat_by | pre-market | Target: zero inventory before primary price discovery resumes and gaps your position. |
overnight_haircut | 0.5 | Quote size multiplier for the overnight and weekend regimes. |
Unwinding without paying the spread back
The elegant unwind is passive: skew until the market fills you back to flat, so you
collect the spread even while reducing. The blunt unwind is crossing someone else's
spread, which refunds edge — acceptable when a deadline (the open, a scheduled
release) is worth more than the cost. The kit prefers passive, escalates to blunt as
flat_by approaches.
On hedging
There is no native short on the token pairs, so classic delta-hedging is limited.
Practical substitutes, in order of preference: quote correlated pairs against each
other so inventories offset; hold the haircut small enough that unhedged exposure is
tolerable; and treat flat_by as a hard constraint rather than a wish.
The strategy's promise — "flat by morning" — is a risk decision, not a slogan.