Skip to content

Commit

Permalink
Incremental overround in booksum estimation
Browse files Browse the repository at this point in the history
  • Loading branch information
@ekoutanov
ekoutanov committed Dec 16, 2023
1 parent 2db192d commit ca46102
Show file tree
Hide file tree
Showing 14 changed files with 297 additions and 217 deletions.
8 changes: 5 additions & 3 deletions brumby-soccer/benches/cri_interval.rs
View file
Original file line number Diff line number Diff line change
@@ -1,15 +1,17 @@
use criterion::{criterion_group, criterion_main, Criterion};

use brumby_soccer::interval;
use brumby_soccer::interval::{IntervalConfig, PruneThresholds, ScoringProbs};
use brumby_soccer::interval::{IntervalConfig, PruneThresholds, BivariateProbs, TeamProbs};

fn criterion_benchmark(c: &mut Criterion) {
fn run(intervals: u8, max_total_goals: u16) -> usize {
interval::explore(
&IntervalConfig {
intervals,
h1_probs: ScoringProbs { home_prob: 0.25, away_prob: 0.25, common_prob: 0.25 },
h2_probs: ScoringProbs { home_prob: 0.25, away_prob: 0.25, common_prob: 0.25 },
team_probs: TeamProbs {
h1_goals: BivariateProbs { home: 0.25, away: 0.25, common: 0.25 },
h2_goals: BivariateProbs { home: 0.25, away: 0.25, common: 0.25 },
},
player_probs: vec![],
prune_thresholds: PruneThresholds {
max_total_goals,
Expand Down
8 changes: 5 additions & 3 deletions brumby-soccer/benches/cri_isolate.rs
View file
Original file line number Diff line number Diff line change
@@ -1,7 +1,7 @@
use brumby_soccer::domain::{OfferType, OutcomeType, Player, Side};
use criterion::{criterion_group, criterion_main, Criterion};

use brumby_soccer::interval::{explore, Exploration, IntervalConfig, ScoringProbs, PruneThresholds, PlayerProbs};
use brumby_soccer::interval::{explore, Exploration, IntervalConfig, BivariateProbs, PruneThresholds, PlayerProbs, TeamProbs};
use brumby_soccer::interval::query::isolate;

fn criterion_benchmark(c: &mut Criterion) {
Expand All @@ -10,8 +10,10 @@ fn criterion_benchmark(c: &mut Criterion) {
explore(
&IntervalConfig {
intervals,
h1_probs: ScoringProbs { home_prob: 0.25, away_prob: 0.25, common_prob: 0.25 },
h2_probs: ScoringProbs { home_prob: 0.25, away_prob: 0.25, common_prob: 0.25 },
team_probs: TeamProbs {
h1_goals: BivariateProbs { home: 0.25, away: 0.25, common: 0.25 },
h2_goals: BivariateProbs { home: 0.25, away: 0.25, common: 0.25 },
},
player_probs: vec![(player, PlayerProbs { goal: Some(0.25), assist: None })],
prune_thresholds: PruneThresholds {
max_total_goals,
Expand Down
137 changes: 89 additions & 48 deletions brumby-soccer/src/bin/soc_prices.rs
View file
Original file line number Diff line number Diff line change
Expand Up @@ -5,27 +5,28 @@ use std::path::PathBuf;

use anyhow::bail;
use clap::Parser;
use racing_scraper::sports::Provider;
use stanza::renderer::console::Console;
use stanza::renderer::Renderer;
use tracing::{debug, info};
use brumby::feed_id::FeedId;

use brumby::hash_lookup::HashLookup;
use brumby::market::{Market, Overround, OverroundMethod, PriceBounds};
use brumby::probs::SliceExt;
use brumby_soccer::{fit, print};
use brumby_soccer::data::{ContestSummary, DataProvider, download_by_id, SoccerFeedId};
use brumby_soccer::data::{download_by_id, ContestSummary, SoccerFeedId};
use brumby_soccer::domain::{
FittingErrors, Offer, OfferType, OutcomeType, Over, Period, Score,
FittingErrors, Offer, OfferType, OutcomeType, Over, Period, Player, Score, Side,
};
use brumby_soccer::fit::ErrorType;
use brumby_soccer::interval::{Expansions, Exploration, explore, IntervalConfig, PlayerProbs, PruneThresholds, ScoringProbs};
use brumby_soccer::interval::query::isolate;
use brumby_soccer::interval::{
explore, BivariateProbs, Expansions, Exploration, IntervalConfig, PlayerProbs, PruneThresholds,
TeamProbs,
};
use brumby_soccer::{fit, print};

const OVERROUND_METHOD: OverroundMethod = OverroundMethod::OddsRatio;
const SINGLE_PRICE_BOUNDS: PriceBounds = 1.01..=301.0;
const FIRST_GOALSCORER_BOOKSUM: f64 = 1.9;
const FIRST_GOALSCORER_BOOKSUM: f64 = 1.5;
const INTERVALS: usize = 18;
// const MAX_TOTAL_GOALS_HALF: u16 = 4;
const MAX_TOTAL_GOALS_FULL: u16 = 8;
Expand Down Expand Up @@ -152,8 +153,8 @@ async fn main() -> Result<(), Box<dyn Error>> {
// h2_search_outcome.optimal_values.normalise(ft_gamma_sum * 1.0);

let exploration = explore_scores(
ScoringProbs::from(adj_optimal_h1.as_slice()),
ScoringProbs::from(adj_optimal_h2.as_slice()),
BivariateProbs::from(adj_optimal_h1.as_slice()),
BivariateProbs::from(adj_optimal_h2.as_slice()),
);

// let mut ft_scoregrid = allocate_scoregrid(MAX_TOTAL_GOALS_FULL);
Expand Down Expand Up @@ -385,6 +386,19 @@ async fn main() -> Result<(), Box<dyn Error>> {
FIRST_GOALSCORER_BOOKSUM,
);
let anytime_gs = fit_offer(OfferType::AnytimeGoalscorer, &anytime_gs, 1.0);
let home_goalscorer_booksum = first_gs
.market
.probs
.iter()
.zip(first_gs.outcomes.items().iter())
.filter(|(_, outcome_type)| {
matches!(
outcome_type,
OutcomeType::Player(Player::Named(Side::Home, _))
)
})
.map(|(prob, _)| prob)
.sum::<f64>();

// println!("scoregrid:\n{}sum: {}", scoregrid.verbose(), scoregrid.flatten().sum());
let draw_prob = isolate(
Expand Down Expand Up @@ -434,8 +448,8 @@ async fn main() -> Result<(), Box<dyn Error>> {
// let elapsed = start.elapsed();
// println!("player fitting took {elapsed:?}");
let fitted_goalscorer_probs = fit::fit_first_goalscorer_all(
&ScoringProbs::from(adj_optimal_h1.as_slice()),
&ScoringProbs::from(adj_optimal_h2.as_slice()),
&BivariateProbs::from(adj_optimal_h1.as_slice()),
&BivariateProbs::from(adj_optimal_h2.as_slice()),
&first_gs,
draw_prob,
);
Expand All @@ -445,11 +459,17 @@ async fn main() -> Result<(), Box<dyn Error>> {
let exploration = explore(
&IntervalConfig {
intervals: INTERVALS as u8,
// h1_probs: ModelParams { home_prob: ft_search_outcome.optimal_values[0], away_prob: ft_search_outcome.optimal_values[1], common_prob: ft_search_outcome.optimal_values[2] },
// h2_probs: ModelParams { home_prob: ft_search_outcome.optimal_values[0], away_prob: ft_search_outcome.optimal_values[1], common_prob: ft_search_outcome.optimal_values[2] },
h1_probs: ScoringProbs::from(adj_optimal_h1.as_slice()),
h2_probs: ScoringProbs::from(adj_optimal_h2.as_slice()),
player_probs: vec![(player.clone(), PlayerProbs { goal: Some(*prob), assist: None })],
team_probs: TeamProbs {
h1_goals: BivariateProbs::from(adj_optimal_h1.as_slice()),
h2_goals: BivariateProbs::from(adj_optimal_h2.as_slice()),
},
player_probs: vec![(
player.clone(),
PlayerProbs {
goal: Some(*prob),
assist: None,
},
)],
prune_thresholds: PruneThresholds {
max_total_goals: MAX_TOTAL_GOALS_FULL,
min_prob: GOALSCORER_MIN_PROB,
Expand All @@ -474,9 +494,10 @@ async fn main() -> Result<(), Box<dyn Error>> {
fitted_first_goalscorer_probs.push(isolated_prob);
// println!("first scorer {player:?}, prob: {isolated_prob:.3}");
}
fitted_first_goalscorer_probs.push(draw_prob);
fitted_first_goalscorer_probs.normalise(FIRST_GOALSCORER_BOOKSUM);
// fitted_first_goalscorer_probs.push(1.0 - fitted_first_goalscorer_probs.sum());

// fitted_first_goalscorer_probs.push(draw_prob);
// fitted_first_goalscorer_probs.normalise(FIRST_GOALSCORER_BOOKSUM);
fitted_first_goalscorer_probs.push(FIRST_GOALSCORER_BOOKSUM - fitted_first_goalscorer_probs.sum());

let fitted_first_goalscorer = Offer {
offer_type: OfferType::FirstGoalscorer,
Expand All @@ -491,14 +512,14 @@ async fn main() -> Result<(), Box<dyn Error>> {
if args.player_goals {
println!(
"sample first goalscorer σ={:.3}",
implied_booksum(&first_gs.market.prices)
first_gs.market.offered_booksum(),
);
let table_first_goalscorer = print::tabulate_offer(&fitted_first_goalscorer);
println!(
"{:?}: [Σ={:.3}, σ={:.3}, n={}]\n{}",
fitted_first_goalscorer.offer_type,
fitted_first_goalscorer.market.probs.sum(),
implied_booksum(&fitted_first_goalscorer.market.prices),
fitted_first_goalscorer.market.offered_booksum(),
fitted_first_goalscorer.market.probs.len(),
Console::default().render(&table_first_goalscorer)
);
Expand All @@ -512,9 +533,17 @@ async fn main() -> Result<(), Box<dyn Error>> {
let exploration = explore(
&IntervalConfig {
intervals: INTERVALS as u8,
h1_probs: ScoringProbs::from(adj_optimal_h1.as_slice()),
h2_probs: ScoringProbs::from(adj_optimal_h2.as_slice()),
player_probs: vec![(player.clone(), PlayerProbs { goal: Some(*prob), assist: None })],
team_probs: TeamProbs {
h1_goals: BivariateProbs::from(adj_optimal_h1.as_slice()),
h2_goals: BivariateProbs::from(adj_optimal_h2.as_slice()),
},
player_probs: vec![(
player.clone(),
PlayerProbs {
goal: Some(*prob),
assist: None,
},
)],
prune_thresholds: PruneThresholds {
max_total_goals: MAX_TOTAL_GOALS_FULL,
min_prob: GOALSCORER_MIN_PROB,
Expand Down Expand Up @@ -543,7 +572,7 @@ async fn main() -> Result<(), Box<dyn Error>> {

let anytime_goalscorer_overround = Overround {
method: OVERROUND_METHOD,
value: anytime_gs.market.offered_booksum() / fitted_anytime_goalscorer_probs.sum()
value: anytime_gs.market.offered_booksum() / fitted_anytime_goalscorer_probs.sum(),
};
let fitted_anytime_goalscorer = Offer {
offer_type: OfferType::AnytimeGoalscorer,
Expand All @@ -558,46 +587,58 @@ async fn main() -> Result<(), Box<dyn Error>> {
if args.player_goals {
println!(
"sample anytime goalscorer σ={:.3}",
implied_booksum(&anytime_gs.market.prices)
anytime_gs.market.offered_booksum(),
);
let table_anytime_goalscorer = print::tabulate_offer(&fitted_anytime_goalscorer);
println!(
"{:?}: [Σ={:.3}, σ={:.3}, n={}]\n{}",
fitted_anytime_goalscorer.offer_type,
fitted_anytime_goalscorer.market.probs.sum(),
implied_booksum(&fitted_anytime_goalscorer.market.prices),
fitted_anytime_goalscorer.market.offered_booksum(),
fitted_anytime_goalscorer.market.probs.len(),
Console::default().render(&table_anytime_goalscorer)
);
}

let sample_anytime_assist_booksum = anytime_assist.values().map(|price| 1.0 / price).sum::<f64>();
let sample_anytime_assist_booksum = anytime_assist
.values()
.map(|price| 1.0 / price)
.sum::<f64>();

let per_outcome_overround = (anytime_goalscorer_overround.value - 1.0) / anytime_gs.outcomes.len() as f64;

let anytime_assist = fit_offer(
OfferType::AnytimeAssist,
&anytime_assist,
sample_anytime_assist_booksum / anytime_goalscorer_overround.value,
sample_anytime_assist_booksum / (1.0 + per_outcome_overround * anytime_assist.len() as f64)
);

let fitted_assist_probs = fit::fit_anytime_assist_all(
&ScoringProbs::from(adj_optimal_h1.as_slice()),
&ScoringProbs::from(adj_optimal_h2.as_slice()),
&BivariateProbs::from(adj_optimal_h1.as_slice()),
&BivariateProbs::from(adj_optimal_h2.as_slice()),
&anytime_assist,
draw_prob,
anytime_assist.market.fair_booksum()
anytime_assist.market.fair_booksum(),
);

let mut fitted_anytime_assist_outcomes =
HashLookup::with_capacity(fitted_assist_probs.len());
let mut fitted_anytime_assist_outcomes = HashLookup::with_capacity(fitted_assist_probs.len());
let mut fitted_anytime_assist_probs = Vec::with_capacity(fitted_assist_probs.len());
for (player, prob) in &fitted_assist_probs {
fitted_anytime_assist_outcomes.push(OutcomeType::Player(player.clone()));
let exploration = explore(
&IntervalConfig {
intervals: INTERVALS as u8,
h1_probs: ScoringProbs::from(adj_optimal_h1.as_slice()),
h2_probs: ScoringProbs::from(adj_optimal_h2.as_slice()),
player_probs: vec![(player.clone(), PlayerProbs { goal: None, assist: Some(*prob) })],
team_probs: TeamProbs {
h1_goals: BivariateProbs::from(adj_optimal_h1.as_slice()),
h2_goals: BivariateProbs::from(adj_optimal_h2.as_slice()),
},
player_probs: vec![(
player.clone(),
PlayerProbs {
goal: None,
assist: Some(*prob),
},
)],
prune_thresholds: PruneThresholds {
max_total_goals: MAX_TOTAL_GOALS_FULL,
min_prob: GOALSCORER_MIN_PROB,
Expand All @@ -621,13 +662,14 @@ async fn main() -> Result<(), Box<dyn Error>> {
);
fitted_anytime_assist_probs.push(isolated_prob);
}
fitted_anytime_assist_outcomes.push(OutcomeType::None);
fitted_anytime_assist_probs.push(draw_prob);

let anytime_assist_overround = Overround {
method: OVERROUND_METHOD,
value: anytime_assist.market.offered_booksum() / fitted_anytime_assist_probs.sum()
value: anytime_assist.market.offered_booksum() / fitted_anytime_assist_probs.sum(),
};
fitted_anytime_assist_outcomes.push(OutcomeType::None);
fitted_anytime_assist_probs.push(draw_prob);

let fitted_anytime_assist = Offer {
offer_type: OfferType::AnytimeAssist,
outcomes: fitted_anytime_assist_outcomes,
Expand All @@ -641,14 +683,14 @@ async fn main() -> Result<(), Box<dyn Error>> {
if args.player_assists {
println!(
"sample anytime assists σ={:.3}",
implied_booksum(&anytime_assist.market.prices)
anytime_assist.market.offered_booksum(),
);
let table_anytime_assist = print::tabulate_offer(&fitted_anytime_assist);
println!(
"{:?}: [Σ={:.3}, σ={:.3}, n={}]\n{}",
fitted_anytime_assist.offer_type,
fitted_anytime_assist.market.probs.sum(),
implied_booksum(&fitted_anytime_assist.market.prices),
fitted_anytime_assist.market.offered_booksum(),
fitted_anytime_assist.market.probs.len(),
Console::default().render(&table_anytime_assist)
);
Expand Down Expand Up @@ -747,9 +789,9 @@ async fn main() -> Result<(), Box<dyn Error>> {
Ok(())
}

fn implied_booksum(prices: &[f64]) -> f64 {
prices.invert().sum()
}
// fn implied_booksum(prices: &[f64]) -> f64 {
// prices.invert().sum()
// }

fn fit_offer(offer_type: OfferType, map: &HashMap<OutcomeType, f64>, normal: f64) -> Offer {
let mut entries = map.iter().collect::<Vec<_>>();
Expand Down Expand Up @@ -935,12 +977,11 @@ fn fit_offer(offer_type: OfferType, map: &HashMap<OutcomeType, f64>, normal: f64
// // scoregrid::inflate_zero(ZERO_INFLATION, &mut scoregrid);
// }

fn explore_scores(h1_probs: ScoringProbs, h2_probs: ScoringProbs) -> Exploration {
fn explore_scores(h1_goals: BivariateProbs, h2_goals: BivariateProbs) -> Exploration {
explore(
&IntervalConfig {
intervals: INTERVALS as u8,
h1_probs,
h2_probs,
team_probs: TeamProbs { h1_goals, h2_goals },
player_probs: vec![],
prune_thresholds: PruneThresholds {
max_total_goals: MAX_TOTAL_GOALS_FULL,
Expand Down
Loading

0 comments on commit ca46102

Please sign in to comment.