[世界杯]根据赔率计算各种组合可能性与赔率

一、背景

本文以世界杯体彩“混合过关”4场串胜平负为的赔率进行编码

其他类型如比分 、总进球数可以参考代码进行相应修改

需要的库：numpy与pandas

二、数据输入

2.1 赔率示意图

2.2 字典保存数据

采用字典保存各比赛对应的胜平负的赔率

import pandas as pd
import numpy as np

vs1 = {"胜":2.38,"平":2.93,"负":2.65} #厄瓜多尔-塞内加尔
vs2 = {"胜":13.0,"平":6.20,"负":1.11} #卡塔尔-荷兰
vs3 = {"胜":3.58,"平":3.16,"负":1.84} #伊朗-美国
vs4 = {"胜":7.35,"平":4.16,"负":1.31} #威尔士-英格兰

三、数据处理

3.1 计算各种组合可能性

计算采用的公式主要为：（图中10%为抽水率,仅为假设）

其中0.9913为初步计算得到的体彩抽水率，实际不准确，该数值仅供初步计算，之后需要根据计算所得的概率进行相应修正。

count = 1
probList = []
probListIndex = []
probVs = []
timesList = []

vsCode1 = []
vsCode2 = []
vsCode3 = []
vsCode4 = []

for key1,each1 in vs1.items():
    for key2,each2 in vs2.items():
        for key3,each3 in vs3.items():
            for key4,each4 in vs4.items():
#             print(count,key1,key2,key3,each1*each2*each3)
            
                prob = 99.13 / (each1*each2*each3*each4)
                #print(count,key1,key2,key3,key4,prob)
                probList.append(prob)
                probListIndex.append(count)
                probVs.append(key1+key2+key3+key4)
                timesList.append(each1*each2*each3*each4)
                vsCode1.append(key1)
                vsCode2.append(key2)
                vsCode3.append(key3)
                vsCode4.append(key4)

                count += 1

data = pd.DataFrame(probList,index = probListIndex,columns=["prob"])
data_temp = pd.DataFrame(probVs,index = probListIndex,columns=["vs"])
data_times = pd.DataFrame(timesList,index = probListIndex,columns=["times"])

data_vs1 = pd.DataFrame(vsCode1,index = probListIndex,columns=["vs1"])
data_vs2 = pd.DataFrame(vsCode2,index = probListIndex,columns=["vs2"])
data_vs3 = pd.DataFrame(vsCode3,index = probListIndex,columns=["vs3"])
data_vs4 = pd.DataFrame(vsCode4,index = probListIndex,columns=["vs4"])

# data = data.add(data_temp,fill_value=False)
data["vs"] = data_temp["vs"]
data["vs1"] = data_vs1["vs1"]
data["vs2"] = data_vs2["vs2"]
data["vs3"] = data_vs3["vs3"]
data["vs4"] = data_vs4["vs4"]

data["times"] = data_times["times"]
data = data.sort_values(by="prob",ascending=False)
data["total_prob"] = 0

sum_prob = 0
for each in data.index:
#     print(each)
    sum_prob += data["prob"].loc[each]
    data["total_prob"].loc[each] = sum_prob
    

3.2 修正概率

该段代码无实际含义，仅为修正由于采用估计抽水率计算所得的概率偏差

主要思路是采用数据标准化后并将数据映射到合理的区间，并对部分概率进行转换

total_prob_min = data["total_prob"].min()
data["total_prob"] = (data["total_prob"]-data["total_prob"].min())/(data["total_prob"].max()-data["total_prob"].min())*(100-total_prob_min)+total_prob_min
data["total_prob"].iloc[0] = (data["total_prob"].iloc[1]*data["total_prob"].iloc[0])/(data["prob"].iloc[1]+data["total_prob"].iloc[0])

temp = data["total_prob"] - data["total_prob"].shift(1)
temp[0] = data["total_prob"].iloc[0]
data["prob"] = temp
data["prob"].iloc[0] = data["total_prob"].iloc[0]
data.to_csv(r"C:\Users\kkkk\Desktop\世界杯1129.csv")

四、输出结果

prob该组合可能性，total_prob为累计可能性，times为赔率，VS1~4为该组合对应的胜平负

以11.29日赛程为参考，卡塔尔与威尔士大概率负，因此采用Excel筛选出相关组合，在所列组合中选取赔率较高的组合。

五、代码

import pandas as pd
import numpy as np

vs1 = {"胜":2.38,"平":2.93,"负":2.65}
vs2 = {"胜":13.0,"平":6.20,"负":1.11}
vs3 = {"胜":3.58,"平":3.16,"负":1.84}
vs4 = {"胜":7.35,"平":4.16,"负":1.31}

count = 1
probList = []
probListIndex = []
probVs = []
timesList = []

vsCode1 = []
vsCode2 = []
vsCode3 = []
vsCode4 = []

for key1,each1 in vs1.items():
    for key2,each2 in vs2.items():
        for key3,each3 in vs3.items():
            for key4,each4 in vs4.items():
#             print(count,key1,key2,key3,each1*each2*each3)
            
                prob = 99.13 / (each1*each2*each3*each4)
                print(count,key1,key2,key3,key4,prob)
                probList.append(prob)
                probListIndex.append(count)
                probVs.append(key1+key2+key3+key4)
                timesList.append(each1*each2*each3*each4)
                vsCode1.append(key1)
                vsCode2.append(key2)
                vsCode3.append(key3)
                vsCode4.append(key4)

                count += 1

data = pd.DataFrame(probList,index = probListIndex,columns=["prob"])
data_temp = pd.DataFrame(probVs,index = probListIndex,columns=["vs"])
data_times = pd.DataFrame(timesList,index = probListIndex,columns=["times"])

data_vs1 = pd.DataFrame(vsCode1,index = probListIndex,columns=["vs1"])
data_vs2 = pd.DataFrame(vsCode2,index = probListIndex,columns=["vs2"])
data_vs3 = pd.DataFrame(vsCode3,index = probListIndex,columns=["vs3"])
data_vs4 = pd.DataFrame(vsCode4,index = probListIndex,columns=["vs4"])

# data = data.add(data_temp,fill_value=False)
data["vs"] = data_temp["vs"]
data["vs1"] = data_vs1["vs1"]
data["vs2"] = data_vs2["vs2"]
data["vs3"] = data_vs3["vs3"]
data["vs4"] = data_vs4["vs4"]

data["times"] = data_times["times"]
data = data.sort_values(by="prob",ascending=False)
data["total_prob"] = 0

sum_prob = 0
for each in data.index:
#     print(each)
    sum_prob += data["prob"].loc[each]
    data["total_prob"].loc[each] = sum_prob
    
total_prob_min = data["total_prob"].min()
data["total_prob"] = (data["total_prob"]-data["total_prob"].min())/(data["total_prob"].max()-data["total_prob"].min())*(100-total_prob_min)+total_prob_min
data["total_prob"].iloc[0] = (data["total_prob"].iloc[1]*data["total_prob"].iloc[0])/(data["prob"].iloc[1]+data["total_prob"].iloc[0])

temp = data["total_prob"] - data["total_prob"].shift(1)
temp[0] = data["total_prob"].iloc[0]
data["prob"] = temp
data["prob"].iloc[0] = data["total_prob"].iloc[0]

data.to_csv(r"C:\Users\kkkk\Desktop\世界杯1129.csv")