Trueskill model in pymc3

Hi experts

I am trying to build a simple Trueskill model in pymc3. My code (draft) is below. My question is how I can build a indicator function in pymc3 that can evaluate:
perf_ player_1>perf_Player_2 then player_1 wins.

winnerdata=np.array([0, 0, 0, 1, 3, 4 ])
loserdata= np.array([1, 3, 4, 2, 1, 2 ])
#Numbers of games
K=len(winnerdata)

#Number of players
#games_matrix=np.append(winnerdata,loserdata)
Games=(games_matrix).max()+1

with pm.Model() as model:
skill_k=pm.Normal(‘skill’,mu=6,sigma=9,shape=K, testval=np.random.randn(K))
perf_k=pm.Normal(‘perf’,mu=skill_k,sigma=9,shape=K)
#perf_k_loose=pm.Normal(‘skill’,mu=skill_k[skill_k[loosedata]],sigma=9,shape=K)
trace = pm.sample(1000, tune=500)

There are a few related posts you can check out: