It’s hard to know which mutual fund is the best for retirement savers to invest in. Each pay period (usually every two or four weeks), employees place another bet on the mutual funds in their 401(k) plans.

Related: 10 tips to boost retirement savings

Without extra help, they usually only have information available from the prospectus on which to make their decision. We all know what they do. They immediate turn to the page that contains the SEC-required performance reporting data. Now, we all know that “past performance does not indicate future results,” but the fact is it can’t be wholly ignored.

That might be a mistake. Here’s why.

I took a look at three of the top 10 most popular retirement plan mutual funds that appeared to have the most similar investment style. These were the Vanguard Institutional Index Fund (VINIX), the Fidelity Contrafund (FCNTX) and the Dodge & Cox Fund (DODGX). Here’s a comparison of their SEC-required performance data (all data based on 12/31/15 figures, see Figure 1 below.)

Based on this performance, there’s no clear winner. DODGX appears to have the greatest recent success, but FCNTX seems to have been better in the long run. On the other hand, VINIX looks like it has done better in the mid-term, as well as finishing in the middle at both the short and long ends of the spectrum.

This, together with an apparently attractive low expense ratio (though we all know it’s irrelevant given we are looking at net of expense performance numbers), might tempt the retirement saver (and many a prudent fiduciary) to choose VINIX.

Related: Think you’re set for retirement? Don’t get too comfortable

This, however, could be a very damaging decision. The Snapshot-in-Time Anomaly shows us why.

The SEC-required performance reporting assumes one invests the money in a lump-sum at the beginning of the period. We all know this is not the case for 401(k) plans. The money is continually invested over time. To better reflect this reality, I took a look at rolling five-year periods over a month-to-month basis. In addition, I looked at a range of goal-oriented-targets (GOTs) for each fund. Given the intermediate term success we saw in VINIX, one might think this would give an unfair advantage to that Vanguard fund. The comparison proves otherwise. Figure two (right) indicates how many periods each fund met or exceeded the particular GOT (i.e., the fund’s “performance frequency” for each GOT).

From this table, with the exception of very low GOTs, it looks like DODGX is the best investment. We might interpret this, together with the SEC-required performance data, as meaning that while DODGX is more likely to meet or exceed the GOT, when it is bad, it is very, very, bad.

But we don’t have to guess. We can actually test this by analyzing the actual performance data of the rolling five year periods to determine how much the highs offset the lows. From this we can calculate the likelihood the fund will break even over time at any particular GOT (i.e., the “performance amplitude” for each GOT).

The performance amplitude (see Figure 3, above) suggests those with lower GOTs (less than 8 percent) might be better served by FCNTX while those with very high GOTs (8 percent-10 percent) might be better served by DODGX. In either case, it’s clear that VINIX is the worst choice. It’s just another fine mess regulators have gotten us into.