Skip to main content
POST
Run tournament simulation

Authorizations

Authorization
string
header
required

Bearer token issued by Softmax

Path Parameters

division_id
string
required

Division whose recorded results should be simulated.

Pattern: ^div_[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}$

Body

application/json
structure
enum<string>
required

Tournament structure to simulate.

Available options:
single_elim,
double_elim,
round_robin,
elimination_rounds,
evolutionary
rounds_window
integer
default:20

Recent completed rounds used when no explicit time window is set.

Required range: 1 <= x <= 200
window_start
string<date-time> | null

Earliest round creation time to include. Overrides rounds_window and retains at most 200 rounds.

window_end
string<date-time> | null

Latest round creation time to include.

coworld_version_min
string | null
default:latest

Oldest Coworld version to include; latest selects the newest version and null removes the bound.

coworld_version_max
string | null

Newest Coworld version to include; latest selects the newest version in the window.

n_sims
integer
default:1000

Monte Carlo tournament runs to sample.

Required range: 1 <= x <= 2000
seed
integer
default:0

Random seed for deterministic simulation results.

best_of
integer
default:1

Odd number of games in each matchup series.

Required range: 1 <= x <= 15
seeding
enum<string>
default:record

How entrants are assigned bracket seeds.

Available options:
record,
random
mutual_rule
enum<string>
default:loss_both

How recorded mutual-loss outcomes are resolved.

Available options:
coinflip,
redraw,
loss_both
smoothing
number
default:1

Additive smoothing applied to pairwise win rates.

Required range: 0 <= x <= 10
rr_passes
integer
default:1

Round-robin passes over every entrant pairing.

Required range: 1 <= x <= 20
eliminate_per_round
integer
default:1

Lowest-ranked entrants removed after each elimination round.

Required range: 1 <= x <= 8
grand_final_reset
boolean
default:true

Play a reset final when the losers-bracket winner wins the first grand final.

pop_size
integer
default:100

Population size for evolutionary tournaments.

Required range: 2 <= x <= 500
generations
integer
default:200

Generations run for evolutionary tournaments.

Required range: 1 <= x <= 500
roster_mode
enum<string>
default:window

Roster selection rule when policy_version_ids is omitted.

Available options:
window,
champions,
top_n,
top_percent
roster_n
integer
default:8

Entrants selected by top_n roster mode.

Required range: 2 <= x <= 32
roster_percent
number
default:50

Percent of entrants selected by top_percent roster mode.

Required range: x <= 100
roster_only
boolean
default:false

Return the roster and win model without running simulations.

sufficiency
boolean
default:true

Estimate whether the recorded pairwise data is sufficient for stable placements.

sufficiency_draws
integer
default:24

Posterior win-model samples for sufficiency analysis.

Required range: 1 <= x <= 100
sufficiency_inner_sims
integer
default:150

Tournament runs evaluated for each sufficiency sample.

Required range: 10 <= x <= 500
policy_version_ids
string[] | null

Explicit policy-version roster; takes precedence over roster_mode.

exclude_policy_version_ids
string[]

Policy versions removed after roster selection.

policy_cutoff
string<date-time> | null

Exclude later submissions and resolve each player's champion as of this time.

Response

Successful Response

division_id
string
required

Division whose recorded results were simulated.

league_id
string
required

League containing the division.

matrix
Matrix · object
required

Pairwise result counts and estimated win rates.

roster
Roster · object[]
required

Policies selected for the simulation with their window records.

note_prefix
string
default:""

Prefix used to identify data requests scheduled for this simulator.

missing_data_pairs
Missing Data Pairs · object[]

Roster policy pairs with no recorded result in the selected window.

data_requests
Data Requests · object[]

Existing experience requests scheduled to fill simulator data gaps.

access
Access · object | null

Caller access and league visibility details.

coworld_versions
Coworld Versions · object[]

Recent Coworld versions available for scheduling more episodes.

win_model
Win Model · object | null

Smoothed pairwise probabilities used by the simulator.

config
Config · object | null

Normalized tournament configuration used for this run.

showcase
Showcase · object | null

One deterministic example tournament run.

monte_carlo
Monte Carlo · object | null

Placement distributions across sampled runs.

sufficiency
Sufficiency · object | null

Stability analysis for the recorded pairwise data.