Analysis Engines

I used to use Stockfish 7 and Komodo 9 for all analysis. But recently, I discovered something. Just for fun, I switched on Fritz 14. Immediately, I noticed that it was recommending moves that I could actually relate to. What I'm saying is, in a lot of positions, especially middle game with equal evaluation, Stockfish will show moves that basically don't do anything other than shuffle a piece to a different, perfectly good square, since it knows the position is drawn. However, engines like Fritz, and Texel 1.05, which I also now use, will make a threatening move on a potential weakness. Even though nothing can come of it with correct play. What I'm saying is, the slightly lower rated (although still beyond 2800), engines, are a more practical choice if your looking for an actual realistic winning chance type of move. Any other opinions or approaches?

I am not sure I understand what you are saying. If you are only doing post-game analysis, and not using the engine to pick moves with "actual realistic winning chances" in ongoing games,  what's the point of learning these provocative suggestions?

I don't know for sure, but it might have something to do with the fact that the default contempt value for Stockfish is set to zero.

Maybe put some effort into figuring out why an engine suggested a move instead of dismissing it as random shuffling.  Komodo makes plenty of human moves that are easy to understand.

I can relate the OP's experience as well. I donate a lot of idle CPU cycles to Chessbase's Let's Check, with a variety of engines: the latest SF dev builds, v9.x Komodo, Houdini 4 Pro (regular and Tactical) and Gull for the most part. For a given position the Let's Check db gives you the first three plies of the PV plus the eval; up to three PVs from three different engines per position.

After seeing the engines chug through hundreds of positions I observed that the top engines usually produce the same PVs (sometimes with transpositions) and very similar evals. That's probably part of the reason that older engines (like the Fritz family, SF5, Houdini 3 and even Rybka 4.1) are still quite popular there - they often provide a second opinion, so to speak.

Of these, Deep Fritz 14 is among the most commonly used. It actually has nothing in common with the original Fritz lineup (which was developed by the same programmers until v12, IIRC) besides the name, it's really a private engine Pandix, adopted by Chessbase to continue the Fritz lineup. This engine does often suggest more "human" moves that the two top dogs, and the curious thing is that sometimes if you let SF run till say d=34-36 it will agree with DF14's PV and eval produced at a much lower iteration count (say, d=22-24).

Texel is another interesting engine (it's up to v1.06a45 at the moment). I've used it a bit on Let's Check but found the fact that its evals very often differ from SF and K's by about half a centipawn a bit disconcerting. Perhaps it's not ready for prime time yet.

I wouldn't use Texel for game analysis although it has a neat swindle function in endgame positions if you have both Syzygy and Gaviota tablebases installed.

There are some good purpose-built engines like Sting SF or Stockfish matefinder than you can use for very tactical positions if you're looking for an another opinion.

Most people should be fine with SF and Komodo though.