You're trying to define the stronger explosive by comparing the reagents, but in this case comparing the products is far more important.
Trinitrotoluene is a rather oxygen-poor explosive, to the point that when detonated as a pure substance, it produces carbon soot and hydrogen gas:
$$\ce{2C7H5N3O6 -> 3N2 + 5H2 + 12CO + 2C}$$
The production of $\ce{CO}$, $\ce{CO2}$ and $\ce{H2O}$ is highly exothermic, so a lot of extra energy could be obtained if there were more oxygen to burn the soot and hydrogen. However, in detonations there's not nearly enough time for atmospheric oxygen to partake extensively in the reaction. This means that a fair amount of the explosive potential of TNT is wasted, decreasing its detonation yield per gram of substance. One way to compensate this yield loss in oxygen-poor explosives is to mix them with an oxidiser, usually an oxygen-rich substance such as $\ce{KClO3}$.
Now, on to picric acid. The structure and molar mass are very similar to TNT, but replacing the methyl group with a hydroxyl means the oxygen content of picric acid is higher. While still overall oxygen-poor, the detonation proceeds with much less soot and hydrogen generation, meaning a more complete burn, a higher energy release, and consequently a greater explosive yield.
$$\ce{2C6H3N3O7 -> 3N2 + H2 + 2H2O + 12 CO}$$
In reality, the decomposition reactions are more complex. The balanced equations above follow a rule of thumb presented here. However, a more thorough analysis should agree with this simplified result.