Note that Kelly is related to winning ratio p and WL. So practically in trading, commission + tax is a kind of losing money. Losing to other player(trader) is almost same as losing to government(spending to other people) and house(broker), in personal account. In conclusion, do NOT trade too often. Many players lost more to gov+house than to other participant. PS) In many casino games like blackjack and slot, one fight against houses (casino). Maybe there are only(?) two games (trading and poker) to fight against other participants. PS) If one has a record for a few years to figure out p and WL based on his personal logic of trading, then think about life-time comm+tax amount. For example, if you pay comm of $20 (ten roundtrips in IB) everyday, then 240*20=4800=4.8K /year and for the next 50 years(from 30 to 80) it adds up to 240K. The amount of 240K shall be compounded over the 50 years, to bigger than your house price.
Kelly works in theory but in practice there are guys ready to put the squeeze on you when they realize that you have large exposure. A couple of hits like that and Kelly puts you out of business. When I see article about Kelly I skip it and I move to the next. I believe the most successful method is to trade less frequently with small risk in several non-correlated markets.
Kelly sizing doesn't necessarily produce large exposure, it depends on your trade return distribution. The main problem is sucky Kelly formulae (aka sucky Kelly estimators) that tend to oversize. Seriously what is known about Kelly in the public domain is almost criminal. Thorp knows what he advocated is a crappy estimator. Why he did it is anybody's guess.