Pandas adapter for the Betfair API-NG implementation in the Python programming language
betfair_pandas uses Joshua Carps' betfair python implementation - https://github.com/jmcarp/betfair.py which is the reference implementation on https://developer.betfair.com/
Install using GIT
Checkout v0.1.3 for betfair.py, as this is what betfair pandas is based on for now. We require this specific version as we have to patch the betfair.py
git clone -b v0.1.3 https://github.com/jmcarp/betfair.py.git
git clone https://github.com/cdtait/betfair_pandas.gitA patch is needed to fix some betfair.py issues
cd betfair.py
patch -p1 -i ../betfair_pandas/betfair.py-json-datetime-2-1.patchNow install betfair.py
python setup.py installand install betfair_pandas
cd ../betfair_pandas
python setup.py installThere is a samples directory which show the combination of betfair.py with pandas features
Each sample will need to have user,password,ssl certificate and app key set
Import betfair_pandas and betfair plus some support for datetime
from betfair.models import MarketFilter
from betfair import Betfair
import betfair_pandas as bp
import datetime
import dateutil
import pandas as pdLogin to betfair, so you will need to add the varibles app_key,cert_file,username,password with valid values
# ssologin
# To use this you will need app_key,cert_file,username,password
client=Betfair(app_key,cert_file)
client.login(username,password)This is the first betfair_pandas call we make. We pass the client as a parameter instead of calling the the client object. The interface is identical apart from that. It will return panadas DataFrame rather tahn a model object
# List horse racing event ids
event_types=bp.list_event_types(client,filter={'textQuery':"Horse Racing"})Now we will use the id from event_types to create a market filter and then call on the catalogue to list the markets.In this case we set up a filter to ask for the first 5 horse races with WIN market in the UK from now
country_code='GB'
marketFilter={'eventTypeIds':[event_types.id[0]],
'marketCountries':[country_code],
'marketTypeCodes':["WIN"],
'marketStartTime':{'from':datetime.datetime.now()}}
# First 5 horse races, win market, from now
races=bp.list_market_catalogue(client,
filter=marketFilter,
market_projection=['COMPETITION','EVENT','EVENT_TYPE','MARKET_DESCRIPTION',
'RUNNER_DESCRIPTION','MARKET_START_TIME'],
sort="FIRST_TO_START",
max_results=5
)We have this set of races so extract some of the details. This demonstrates that the use of a DataFrame with selecting columns and merging DataFrame's together.
# Get a summary set of columns for winHorseRacing from description
summaryDesc=races['description'][['marketId','marketName','event.venue',
'event.name','marketStartTime']]
# Get a summary set of the runners names
summaryRunners=races['runners'][['marketId','selectionId','runnerName']]
# Join the 2 together baes on the marketId to show summary of the runners in the races together
summaryRaces=pd.merge(summaryDesc,summaryRunners,on='marketId')Now show groupby and printout what we find.
summaryDesc.groupby(['marketStartTime','event.venue'])
print('Races:')
for name, group in summaryDesc.groupby(['marketStartTime','event.venue']):
print("{0:s} {1:s} {2:s} {3:%I:%M%p}".format(group.values[0][1],
group.values[0][2],
group.values[0][3],
dateutil.parser.parse(group.values[0][4])
))Output of group printout
Races:
2m Hcap Chs Fakenham Fake 18th Nov 03:30PM
7f Hcap Southwell Sthl 18th Nov 03:40PM
2m NHF Doncaster Donc 18th Nov 03:50PM
1m Nursery Lingfield Ling 19th Nov 12:00PM
2m Mdn Hrd Hexham Hex 19th Nov 12:20PMSelect the first race, they are order in terms of time so get marketId from row 0
# First race
marketId=races['description'].marketId[0]We use this now to construct another output for details of the first race, this demonstartes the query method on a DataFrame
# First race summary
firstRaceDesc=summaryDesc.query('marketId=="'+marketId+'"')[['marketId','marketName',
'event.name','event.venue','marketStartTime']]
firstRaceRunners=summaryRaces.query('marketId=="'+marketId+'"')[['selectionId','runnerName']]
print(firstRaceDesc)
print(firstRaceRunners)Output of first race description
marketId marketName event.name event.venue marketStartTime
0 1.116413138 2m Hcap Chs Fake 18th Nov Fakenham 2014-11-18T15:30:00.000ZOutput to display runners in first race
selectionId runnerName
0 5718959 Carobello
1 7572682 Larteta
2 4363561 Dynamic Idol
3 4815491 Full Ov BeansGet hold of the prices. This call uses a more complex filter based on a pure dictionary structure as there is a gotcha when using model objects within model objects. So in this case we get 5 levels of the pricing book for exchange and sp and limit this to rollup to 20
# All exchange and starting prices to a depth of 5 max price of 20
projection={'priceData':['EX_BEST_OFFERS','SP_AVAILABLE','SP_TRADED'],
'virtualise':False,
'exBestOffersOverrides':{'bestPricesDepth':5L,
'rollupModel':"STAKE",
'rollupLimit':20L},
'rolloverStakes':False
}
# Get all the runners/prices book for this market
# According to the projections
runnersPriceInFirstRace=bp.list_market_book(client,
market_ids=[marketId],
price_projection=projection,
order_projection='ALL',
match_projection='ROLLED_UP_BY_PRICE'
)
# Note the book time
priceTime=datetime.datetime.now()
print(runnersPriceInFirstRace['market.book'][['marketId','lastMatchTime',
'totalAvailable','totalMatched',
'numberOfActiveRunners']])Ouput the summary of the book
marketId lastMatchTime totalAvailable totalMatched \
0 1.116413138 2014-11-18T15:27:04.235Z 102182.84 191392.47
numberOfActiveRunners
0 4 This section now looks for the runner with the largest total match as this is what we will extract the book for shown later. Gets the overview of the one particular runner we want. This demonstartes the DataFrame where selection
runnerWithMostTotalMatched=runnersPriceInFirstRace['runners'].sort('totalMatched',
ascending=False)
# This is one particular runner
runnerIdWithMostTotalMatched=runnerWithMostTotalMatched.ix[0,'selectionId']
# Getthe overview price and volume for this selection
runners=runnersPriceInFirstRace['runners']
overview=runners[runners.selectionId==runnerIdWithMostTotalMatched]Demostrate the use of merge on multiple columns to create the runner overview used later
# Getthe overview price and volume for this selected runner
runnerOverview=pd.merge(overview,summaryRaces[['selectionId','marketId','runnerName']],
on=['selectionId','marketId'])
allsp=runnersPriceInFirstRace['runners.sp']
sp=allsp[allsp.selectionId==runnerIdWithMostTotalMatched]
# Show starting price summary
print(sp)Output the starting price summary
farPrice marketId nearPrice selectionId
0 2.32 1.116413138 3 5718959Extract the back stake taken. Her eit would be useful to point out that runnersPriceInFirstRace is a dictionary which flattens out the list lines and levels in the return structure and this is what enables us to extract the data in DatFrame format. So you may find it useful to look at the code or look at data returned
runnersPriceInFirstRace.keys()Get the back stake taken
# Show back stake taken
backStakeTaken=runnersPriceInFirstRace['runners.sp.backStakeTaken']
print(backStakeTaken[backStakeTaken.selectionId==runnerIdWithMostTotalMatched])Output the back stake taken
marketId price selectionId size
0 1.116413138 1.01 5718959 891.27
1 1.116413138 1.80 5718959 186.34
2 1.116413138 3.00 5718959 980.99
3 1.116413138 3.45 5718959 498.04
4 1.116413138 6.00 5718959 5.00Get the lay liability taken
# Show lay liabilty taken
layLiabilityTaken=runnersPriceInFirstRace['runners.sp.layLiabilityTaken']
print(layLiabilityTaken[layLiabilityTaken.selectionId==runnerIdWithMostTotalMatched])Output the lay liability taken
marketId price selectionId size
0 1.116413138 1000.00 5718959 823.10
1 1.116413138 3.20 5718959 15.95
2 1.116413138 2.32 5718959 967.62Extract the lay and back prices. This demonstrates the renaming ability we have with DatFrame's
# Get all lay prices for all the runners in the first race
availableToLay=runnersPriceInFirstRace['runners.ex.availableToLay']
# Get the lay prices for the one with the most total matched
runnerIdWithMostTotalMatchedLayPrices=availableToLay[availableToLay.selectionId == runnerIdWithMostTotalMatched]
# Rename to TotalAvailableToLay
runnerIdWithMostTotalMatchedLayPrices=runnerIdWithMostTotalMatchedLayPrices.rename(
columns={'size': 'LayTotal','price':'LayPrice'})
# Get all back prices for all the runners in the first race
availableToBack=runnersPriceInFirstRace['runners.ex.availableToBack']
# Get the back prices for the one with the most total matched
runnerIdWithMostTotalMatchedBackPrices=availableToBack[availableToBack.selectionId == runnerIdWithMostTotalMatched]
# Rename to TotalAvailableToBack
runnerIdWithMostTotalMatchedBackPrices=runnerIdWithMostTotalMatchedBackPrices.rename(
columns={'size': 'BackTotal', 'price':'BackPrice'})Join the lay and back prices and this will allowus to see what the pricing book or ladder looks like
# Merge the prices by appending to make a price ladder
priceLadder=runnerIdWithMostTotalMatchedBackPrices[['BackTotal','BackPrice']].join(
runnerIdWithMostTotalMatchedLayPrices[['LayPrice','LayTotal']])
print("Market:{0:s} {1:s} {2:s} {3:%I:%M%p}".format(firstRaceDesc['marketName'][0],
firstRaceDesc['event.name'][0],
firstRaceDesc['event.venue'][0],
dateutil.parser.parse(firstRaceDesc['marketStartTime'][0]))
)
print("Runner:{0:s} Total {1:f} ".format(runnerOverview['runnerName'][0],
runnerOverview['totalMatched'][0]))
print('Book at {0:s}'.format(priceTime.isoformat(' ')))
print(priceLadder)Output the book
Market:2m Hcap Chs Fake 18th Nov Fakenham 03:30PM
Runner:Carobello Total 111261.690000
Book at 2014-11-18 15:27:04.563642
BackTotal BackPrice LayPrice LayTotal
0 1075.11 3.00 3.05 625.80
1 458.76 2.98 3.10 538.69
2 1314.45 2.96 3.15 350.37
3 1383.35 2.94 3.20 491.22
4 524.50 2.92 3.25 651.02Logout
client.logout()
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent