todds1 / todds0 / todds2 fields are fair decimal odds, with margin removed using the selected model. They are not probabilities: the corresponding probability is 1/todds.
They are present on every snapshot of /api/odds (not only opening/closing), which lets you track how the edge evolves over time.
Choosing the devig model
By default,todds use the MPTO model (margin proportional to odds). Five models are available — pick one per request with ?model=, or set your account preference in the dashboard.
Names, aliases and formulas
Canonical names follow the WinnerOdds calculator. Parameters are case-insensitive. Legacy aliases remain accepted:multiplicative = EM, additive = MPTO, power = LOG.
Notation:
oᵢ = raw decimal odds, qᵢ = 1/oᵢ, S = Σqᵢ, m = S − 1, n = number of mutually exclusive outcomes covering the market. pᵢ is the fair probability; fair odds equal 1/pᵢ. For SHIN, OR and LOG, the parameter is solved numerically so that Σpᵢ = 1.
MPTO is our default model. It subtracts
m/n from each implied probability: pᵢ = 1/oᵢ − m/n. Equivalently, fair odds = n × oᵢ / (n − m × oᵢ). This differs from EM, which multiplies each price by S.
?model= (per-request override) > account preference > MPTO. The model applies to all endpoints carrying todds: /api/snapshot, /api/updated, /api/odds, /api/opening, /api/closing, /api/specials/odds and the real-time /api/stream/event stream.
On the
/api/stream/odds feed, ?model= has no effect: the frame is serialised once for all
subscribers, which rules out a per-client model. Every row there carries true_odds_by_model with
all five models — you pick at read time.Snapshot and updates
On/api/snapshot and /api/updated, todds are recalculated from raw odds for each request. Priority: valid ?model= > valid account preference > MPTO. An unknown value falls back to the account preference or default.
The token stores scope and cursor, not the model. Repeat model=MPTO on every /api/updated request. To switch models, reload /api/snapshot: a delta only returns changed rows, not every old price that needs recalculation.
Three-outcome example
For raw odds2.09 / 3.59 / 3.77, our implementation returns the following values rounded to four decimal places.
SHIN / MPTO
SHIN and MPTO are mathematically identical on a complete two-outcome market with positive margin. They can differ with three or more outcomes; EM and LOG remain distinct methods. Identical rounded values alone do not prove a parameter is ignored.Detecting value (+EV)
Compare a bookmaker’s odds to thetodds. If a book offers odds higher than the corresponding todds, it is a positive expected value opportunity.
Since Pinnacle is the sharpest book, its devig odds are your OTB (Odds To Beat) — the reference for identifying value and computing CLV.
Specials & outrights
/api/specials/odds also includes a todds field on each line. All contestants of a special are devig together (2-way or N-outcome market).
CLV
CLV compares your bet price to the closing price. Fetch the close with/api/closing, then apply the formula:
CLV% = (bet_price / closing_price − 1) × 100 — a bet at 3.50 against a 3.11 close gives +12.5%. Prefer comparing devig prices (todds): the margin skews the calculation.