Leverage/Slippage/Commission Factor Against a Long Only Model

Could a resident Q Guro take a look and provide feedback on this taking into account the following:

In an institutional capacity, there is no leverage factor.

Commissions are close to 5% of retail (even free commission has a slippage factor).

All of the dialogue on Q centers around leverage costs and commissions as the ultimate test of whether a model is "worthy" or not.

However, the Q fund should have an institutional footprint where these costs are not as important when measuring the viability of the underlying model. From what I've been reading (mind you I am only a sideline observer), it appears many good models are developed with the intention of gaining access to Q fund capital, but are being written as if they will be charged margin rates and retail commissions.

Is this perhaps a way for Q to create a built in cushion that results in an increase in absolute returns purely from finance cost and commission savings?

With this algo, assuming zero capital costs, and adjusting data granularity such that trade frequency is increased without having to worry about commissions eating into profits would create something that while not worthy of the Q fund, could be worthy of a hedge fund.

I would love to see variations on this algo with the above assumptions factored in that produce significantly higher returns without worrying about excess volatility. At the hedge fund level, portfolio stops can be used to control downside risk. I haven't figured out how to incorporate a portfolio level stop on Q such that when the portfolio is stopped out and goes to 100% cash (or alternative security), the book is reestablished as if the algo was reset the next day.

Any additional insight would be highly educational for me.

Sorry for the slow reply on this - I had it in my to-reply list and forgot about it.

it definitely is worthwhile to evaluate algorithms with low/no commissions and see how they do. You can evaluate without slippage, but you really need to make sure slippage comes back in when you start considering the capacity of your algorithm. When we evaluate algorithms for the fund we definitely take all of this into consideration.

When we look at algorithms for allocations we use a pyfolio tearsheet of a long backtest. You've provided the long backtest so I ran the tearsheet, attached. As you know, we're not looking for long only algorithms. That said, a good long-only algorithm can often be morphed into a hedged one. Some thoughts:

It has some painful drawdowns. Over the 13 year test it has 5 drawdowns of 20%. That's more drawdown that we'd prefer.
The monthly returns are very inconsistent. Most of the algorithm's success is driven by 5 or 6 "lucky" months. We're looking for algorithms that have a more consistent profile. It's hard to predict if there are more lucky months in this algorithm's future.
It spends long periods of time in a single ETF. That's an exposure to a single name that we wouldn't want to make a large allocation to.
When it is "in" the market, it appears to be in 5 names at a time. That's still higher exposure than we'd like. We'd rather see a max exposure to a single name of 10% max.