Baccarat Long-Streak Discrepancy

So I downloaded Michael Shackleford's computer-generated data set of 25,000 Baccarat Shoes played (at about 80-81 hands per shoe), and after counting all the streaks I notice something I don't understand. In that voluminous set, the Banker's edge prevails for streaks of 1 to 12, as expected, but the Player beats that edge for streaks of 13 to 19. Why is this exactly? I suppose one should simply dismiss it as an anomalous and insignificant sampling (a tiny sliver of 60 hands out of 2,022,072 played), but it's such a stark contrast. If it were more checkered -- the high streaks haphazardously alternating between Banker and Player -- then it might barely pique one's interest, but it's an utter switcheroo! Can any 'mathmagicians' out there educate me in this mystery?

Shackleford Computer Baccarat Simulations
25,000 Shoes
2,022,072 Total Hands
80.88288 Hands per Shoe
Streaks of 1 to 19 Count Below

#----Banker----Player----% of Banker
-1---927,132---903,136----97%
-2---288,521---276,165----96%
-3---114,666---107,096----93%
-4----48,939-----45,121----92%
-5----21,735-----19,422----89%
-6-----9,780-------8,455----86%
-7-----4,367-------3,727----85%
-8-----1,971-------1,642----83%
-9-------882---------725----82%
10-------402---------286----71%
11-------170---------140----82%
12--------73-----------55----71%
Banker streak dominance ends.
13--------25-----------27---108%
14--------10-----------15---150%
15---------3-------------9---300%
16---------3-------------6---200%
17---------1-------------2---200%
18---------0-------------0------0%
19---------0-------------1---200%

Quote: CopperTop

So I downloaded Michael Shackleford's computer-generated data set of 25,000 Baccarat Shoes

It is so much fun playing with Mike's Baccarat data! Many thanx for sharing.

Ummm... Did you ignore ties in determining streaks? So, "B,P,P,T,P,P,B" would show as a streak of 4 for the Player, and not two 2-streak sequences, right?

Gosh, you have 9 more collections of 25,000 Baccarat shoes waiting to be checked. Does your initial data hold up across Mike's total collection of Baccarat data? Or, is your data a garden-variety "statistical anomaly"?

Quote: CopperTop

I suppose one should simply dismiss it as an anomalous and insignificant sampling (a tiny sliver of 60 hands out of 2,022,072 played), but it's such a stark contrast.<snip>Can any 'mathmagicians' out there educate me in this mystery?

No Mystery.

I do not see that over 1 million shoes I ran.
I do not think the Wizard wants to run 1 million shoes and post the data
I could be wrong here.

Banker streaks were KING until 19 in a row.

now at 100 million shoes Banker stayed KING of the streaks until 23 in a row.
better things to do now

small sample size for sure.
not worth even placing the data here
Now to quiet down my Mom
Sally

In some games, or looking at how often unlikely events happen, you do need to run lots of hands. I'm more accustomed to looking at BlackJack but typically 100m shoes are needed to get reasonably close to the calculated House Edge.

#----Banker----Player----% of Banker
-1---927,132---903,136----97%
-2---288,521---276,165----96%
-3---114,666---107,096----93%
-4----48,939-----45,121----92%
-5----21,735-----19,422----89%

Seems like there's so much chop with only 1 win in a row. I'll have to start my progressions after the 2nd win in a row.
288K/927K = 31%
114K/288K = 39.6%
49K/114K = 43%
21K/49K = 43%