From d963a5d96352ece425d76233351d3f9c879dbf30 Mon Sep 17 00:00:00 2001
From: "Steinar H. Gunderson" <sgunderson@bigfoot.com>
Date: Mon, 19 Mar 2012 02:52:47 +0100
Subject: [PATCH] Take standard deviation of estimated mu into account when
 doing MC simulation.

---
 mcwordfeud.cpp | 19 ++++++++++++++-----
 www/index.pl   | 20 ++++++++++----------
 2 files changed, 24 insertions(+), 15 deletions(-)

diff --git a/mcwordfeud.cpp b/mcwordfeud.cpp
index 2bd4a52..b48936d 100644
--- a/mcwordfeud.cpp
+++ b/mcwordfeud.cpp
@@ -14,7 +14,7 @@ using namespace std;
 
 #define MAX_PLAYERS 16
 
-float match_stddev = 70.0f;
+float match_variance = 70.0f * 70.0f;
 
 struct player {
 	int player_index;
@@ -67,12 +67,15 @@ float draw_gaussian(float mu, float stddev)
 int main(int argc, char **argv)
 {
 	int trials = atoi(argv[1]);
+	float match_stddev;
 
 	if (scanf("%f", &match_stddev) != 1) {
 		fprintf(stderr, "Could't read match stddev\n");
 		exit(1);
 	}
 
+	match_variance = match_stddev * match_stddev;
+
 	int num_players;
 	if (scanf("%d", &num_players) != 1) {
 		fprintf(stderr, "Could't read number of players\n");
@@ -87,6 +90,7 @@ int main(int argc, char **argv)
 	vector<string> players;
 	map<string, int> player_map;
 	float ratings[MAX_PLAYERS];
+	float ratings_variance[MAX_PLAYERS];
 	bool has_scores[MAX_PLAYERS][MAX_PLAYERS];
 	for (int pl1 = 0; pl1 < num_players; ++pl1) {
 		for (int pl2 = 0; pl2 < num_players; ++pl2) {
@@ -97,19 +101,23 @@ int main(int argc, char **argv)
 	int scores[MAX_PLAYERS][MAX_PLAYERS];
 	for (int i = 0; i < num_players; ++i) {
 		char buf[256];
-		float rating;
-		int ret = scanf("%s %f", buf, &rating);
+		float rating, rating_stddev;
+		int ret = scanf("%s %f %f", buf, &rating, &rating_stddev);
 		if (ret < 1) {
 			fprintf(stderr, "Couldn't read player %d\n", i);
 			exit(1);
 		}
-		if (ret != 2) {
+		if (ret < 2) {
 			rating = 1500.0f;
 		}
+		if (ret < 3) {
+			rating_stddev = 0.0f;
+		}
 
 		players.push_back(buf);
 		player_map[buf] = i;
 		ratings[i] = rating;
+		ratings_variance[i] = rating_stddev * rating_stddev;
 	}
 
 	for ( ;; ) {
@@ -151,8 +159,9 @@ int main(int argc, char **argv)
 				}
 
 				float mu = ratings[pl1] - ratings[pl2];
+				float stddev = sqrt(match_variance + ratings_variance[pl1] + ratings_variance[pl2]);
 				
-				int score = lrintf(draw_gaussian(mu, match_stddev));
+				int score = lrintf(draw_gaussian(mu, stddev));
 				scores[pl1][pl2] = score;
 				scores[pl2][pl1] = -score;
 			}
diff --git a/www/index.pl b/www/index.pl
index 604edc2..d1943ae 100755
--- a/www/index.pl
+++ b/www/index.pl
@@ -25,6 +25,7 @@ binmode STDOUT, ':utf8';
 
 my %players = ();
 my %ratings = ();
+my %ratings_stddev = ();
 my @matches = ();
 my %parms = ();
 my $match_stddev;
@@ -39,7 +40,7 @@ sub color {
 }
 
 sub make_table {
-	my ($lowest_division, $used_ratings) = @_;
+	my ($lowest_division, $used_ratings, $used_ratings_stddev) = @_;
 
 	print <<"EOF";
 
@@ -58,11 +59,9 @@ EOF
 	printf MCCALC "%d\n", scalar keys %players;
 
 	for my $id (keys %players) {
-		if (exists($used_ratings->{$id}) && defined($used_ratings->{$id})) {
-			printf MCCALC "%s %f\n", $id, $used_ratings->{$id};
-		} else {
-			printf MCCALC "%s %f\n", $id, 1500.0;
-		}
+		my $rating = $used_ratings->{$id} // 1500.0;
+		my $rating_stddev = $used_ratings_stddev->{$id} // $parms{-3};
+		printf MCCALC "%s %f %f\n", $id, $rating, $rating_stddev;
 	}
 
 	for my $match (@matches) {
@@ -80,7 +79,7 @@ EOF
 		chomp;
 		my @x = split /\s+/;
 		my $id = $x[0];
-		my $player = sprintf "%s (%.0f)", $players{$id}, $ratings{$id};
+		my $player = sprintf "%s (%.0f &pm; %.0f)", $players{$id}, ($ratings{$id} // 1500.0), ($ratings_stddev{$id} // $parms{-3});
 		$prob{$player} = [ @x[1..$#x] ];
 	}
 	close MCCALC;
@@ -230,13 +229,14 @@ print <<"EOF";
 EOF
 
 # Get players and ratings
-$q = $dbh->prepare('SELECT fotballdeltagere.id,fotballdeltagere.navn,rating FROM fotballdeltagere JOIN fotballserier ON fotballdeltagere.serie=fotballserier.nr LEFT JOIN ratings ON fotballdeltagere.id=ratings.id WHERE sesong=? AND divisjon=? AND avdeling=?');
+$q = $dbh->prepare('SELECT fotballdeltagere.id,fotballdeltagere.navn,rating,rating_stddev FROM fotballdeltagere JOIN fotballserier ON fotballdeltagere.serie=fotballserier.nr LEFT JOIN ratings ON fotballdeltagere.id=ratings.id WHERE sesong=? AND divisjon=? AND avdeling=?');
 $q->execute($season, $division, $subdivision);
 
 while (my $ref = $q->fetchrow_hashref) {
 	my $id = $ref->{'id'};
 	$players{$id} = sanitize(Encode::decode_utf8($ref->{'navn'}));
 	$ratings{$id} = $ref->{'rating'};
+	$ratings_stddev{$id} = $ref->{'rating_stddev'};
 }
 $q->finish;
 
@@ -259,14 +259,14 @@ while (my $ref = $q->fetchrow_hashref) {
 $q->finish;
 
 my $lowest_division = ($division == $max_division);
-make_table($lowest_division, {});
+make_table($lowest_division, {}, {});
 
 print <<"EOF";
     <p>Under er en variant som tar relativ spillestyrke med i beregningen;
       se <a href="/rating">ratingsiden</a>.</p>
 EOF
 
-make_table($lowest_division, \%ratings);
+make_table($lowest_division, \%ratings, \%ratings_stddev);
 
 print << "EOF";
     </table>
-- 
2.39.2