Is this a Useful Training Method ?

A database is the right tool. Use something like chess position trainer to store moves and play them back.

I think playing whole games: opening, middle game, endgame against Stockfish is better than digging into data bases which only relates to the opening. Databases only supply knowledge of what other people have thought, playing makes you think for yourself.

I don't spend money on databases or something. That's the reason why I use lichess.

@a_pleasant_illusion I agree regarding TCEC; I respectfully disagree regarding whether engines play openings suboptimally under nominal circumstances (I overstated my case earlier) -- I see we agree on this point as well...

It's true engines play openings better today than they did in the 20th century; however commercial chess software comes bundled with opening books to be played on behalf of the engine. Certainly an engine given enough time will play reasonable opening moves; however (I'm rephrasing here) optimal opening moves won't reliably be found without minutes of computation time. So many automated analyses and "play against the machine" games will yield moves less optimal than the Masters database in Lichess' Opening Explorer.

I belabor this point only because I receive so many bug reports about SF not playing the best opening move.

Training won endgames against the engine is useful, playing openings not, because your opponents will not play what the engine plays. Indeed use a database and replay master games. And of course databases dont cost anything. Use Scid + Gorgonians game collection + TWIC

@tpr This is about openings. Other people (especially strong players) have put alot more thought into all those moves. They know the ideas and they've already done all the effort trying to figure things out with help of the engine. So you can try to figure out yourself with only the engine, but why ignore all the ideas top players already came up with?

Of course you will analyse the middlegame with an engine.

Yes, but playing is active learning while database browsing is passive learning. Doing something is more effective than reading about what to do. A boxer uses sparring partners, no box videos. Stockfish is an excellent sparring partner: never blunders. It is like a grandmaster that you can summon whenever you like.

@tpr this is true for middlegames and endgames but not for openings. The opening ideas by GMs usually are of a higher quality than the engine suggestions.

NO I PRACTICE FROM MOVE 5/6 AND GENERALLY PLAY TILL THE MID-MIDDLEGAME.
I'LL DECLARE MYSELF SUCSESSFULL IF I DRAW A GAME WITH WHITE.

TWIC is paid. I mean, real NO money. I'm not old enough to send money through the Internet.

http://theweekinchess.com/twic _is_ free. Please scroll down, you see the download links.

Use a software like http://www.freedownloadmanager.org/ to mass download them all (using the '(*)' operator).

Then put those zips into an empty dir, unzip them using for example http://www.7-zip.org/ (you can use the '*' operator to unzip all zips in a dir).

Then in that dir open a commandline and type:
type *.pgn > merged.pgn

then import that merged.pgn into Scid.

Look at the error messages, open the pgn with a fast editor, for example http://tibleiz.net/code-browser/ and go to these lines (i think it is ctrl+g) and fix the errors or delete the game. usually these are illegal moves or chess 960 games. Start with the last error message, not the first, guess why. you can also copypaste the pgn text of that broken game until the error into Scid to see whats going on.

Finally the pgn should be clean and the import gives no errors. Then you can create a new database in Scid format (or, if you have imported gorgonians game collection before, use that database) and dragdrop the games from the pgn database into the new database.

If you are, for performance reasons, just interested in like, games by players above 2300, you can filter the database first, and then dragdrop it. It will just copy games in the filter.

Then search for duplicates, fix ECO, names, tournaments (but NOT Ratings! it adds stupid brackets around the ratings and breaks some ratings).

Finally compress the database.

You should now have around 3 to 5 million games for free. Voila.