Because the score doesn't use the volatilty in the denominator: it uses 0.02 when the volatility happens to be below that threshold.

I see, besides the 0.02 threshold rule, does the score system consider return itself at all?
current interest rate is +2.5% (0.025) which makes the threshold-adjusted sharpe 1.25, which is actually a decent number.
So I guess 1.25 sharpe should be the minimal that should be considered as contest entry, regardless of the return?
As an extreme example, would an algo with 30% annual return but with just 1.20 sharpe be rejected in the contest?

if sharpe ratio is important criteria, why not just invest in risk-free asset?

Quite.

if sharpe ratio is important criteria, why not just invest in risk-free asset?

The short answer to your question is people want a higher annual return than 3% a year. High net worth individuals are expecting 9% returns a year according to several wealth management reports. The point of the contest is to get higher risk adjusted returns using stocks. Equity market neutral stategies are mostly self financed so it is easier to leverage the small returns of 2% to 5% by 2X to 6X. Plus the cash collateral used to construct the portfolio can collect interest.

One word: Leverage. Prime brokers will let you use their capital for cheap if you’re actually trading, so they can profit from commission and clearing fees. Don’t think they’d be as keen to let anyone use their capital if you’re just locking in the risk free rate for 30 years...

If a strategy returns 4% annually with 2% volatility (2.0 sharpe), Q could realistically leverage that 5x and get 20% annual returns. That’s just from a single Algo too. In a portfolio of uncorrelated strategies, volatility would decrease (approach zero in theory) and sharpe increase (approach infinity in theory), so they could potentially leverage even more.

With the risk free rate you’re kinda ‘locked in’ for the full period, unless you’re happy to take a loss during increasing interest rates.