Keys and method
The core loop
Between your picks, mark who went. Everything else recomputes: survival,
tier cliffs, posture fit, candidates, and the brief for your next turn.
| Key | Action |
any letter | Jumps to the mark box on the Live tab, or search on the Board tab |
Enter | Marks the top match gone |
Shift+Enter | Marks the top match as mine |
Escape | Clears the box |
Ctrl+Z | Undo the last pick |
+ | Records one unknown pick (a pick happened, you do not know who) |
b then click | Bulk mode: one click per player, no confirmation |
Incomplete information, handled honestly
If eleven picks have happened and you only marked five, the guide does not
pretend the other six players are still available. Press +1 unknown for
each pick you could not capture (or let the count drift and correct it once).
Those unknown picks are then distributed across the remaining pool in proportion
to how likely each player was to be taken in that window.
Every survival number then splits into two parts:
P(available at my pick) = P(not already gone unseen) x P(survives from here)
and the guide shows the second number alongside the chance he is already
gone and you have not marked it. When that second number is large, the
headline percentage is not to be trusted, and the Info quality box says so.
Postures, not branches
Four postures are held at once. Each is scored two ways:
- Value from here , expected xVOR of all your remaining picks if you
commit to that posture, computed by simulating each remaining pick against the
live board. This is the decision-relevant number.
- Held , how closely your roster so far already matches that posture.
This is what you are actually doing, whatever you intended.
When the highest-value posture stops being the one you are holding, the guide
says so, names what the new posture commits you to, and prices the difference.
No pre-written branches: the tree is too wide and dies on contact.
Reach prevention
A reach warning on ADP delta alone is useless, because disagreeing with the
market is the whole point of the board. The test used here is different:
It is only a reach if deferring costs nothing.
comparable players remain at his position (within 8 xVOR), AND
P(at least one of them survives to my next pick) is high.
If he is the last of his tier, it is not a reach at any ADP, and the
guide says that explicitly rather than staying quiet. The comparables and the
tier-survival probability are both shown, so you can disagree with the call.
Survival model and its assumption
Draft slot is treated as normally distributed around ADP, with the real
measured standard deviation from Fantasy Football Calculator's 10-team half-PPR
mocks where available, else max(6, 0.18 x ADP). Picks are treated as
independent, which they are not: real drafts run. The position-run counter is the
manual correction, and the landmark bands are deliberately coarse for the same
reason.
Storage
State is kept in this browser via localStorage. If that is
blocked, a red banner appears and you should use Export before refreshing.