Testing How To Test.

It seems like it's a practically weekly basis that someone, somewhere, loudly proclaims that they've broken a format in Magic. I'm sure you know exactly what I mean: whether its via article, video, or the timeless medium of coarsely yelling in card stores, everyone seems to think, at some time or another, their pile of sixty absolutely destroys the accepted best deck in the format, and their smaller pile of fifteen ensures they beat the rest of the contenders. I bring this up somewhat consdescendingly, knowing I've been just as guilty in the past. Too often we think something we've built is the nut high in a given format, and it only gets worse when we back it up with our testing. Results never lie, do they?
Yes, they do. While not as foolish as arguing how your deck theoretically beats another deck ("I just play this 6 mana 5/5 shroud guy and Jund scoops... Err, I mean they can't attack for a few turns but neither can I... Err, I mean they strip my hand with double Blightning or kill all my lands with Ruinblaster before I can cast it... But hey, if I do cast it, I get to look at my top card, so there!), the biggest problem with stating how your "testing" proves with results that your deck is insane lies behind the simple fact that most people, frankly, test wrong.
Loading up MWS and firing off a lot of games with the name "Jund decks only" is certainly better than doing nothing, but not by as much as you think. First, your results are skewed both ways, in favour of your deck because there's no way to ensure that your opponent is playing at the level that justifies your testing in the first place, and against your deck because you don't know the ins and outs of your new concoction just yet.
It's no different in other formats either, Jund is simply my favourite example because it's the current Standard boogeyman. How many people thought they had Faeries' wings permanently clipped? After all, they made an enchantment that killed their guys whenever they attacked, and an elf that pyroclasmed the air for one just by tapping, right? And let's not forget the sheer amount of decks that "smashed affinity" back in those dark times of Magic.
The point is, how you test is just as important as how well your deck tests. If you're testing isn't absolutely relevent at all times, you're wasting your time, as well as that of the friends, readers and family pets that have to listen to your bizarre claims, backed up by your flawed testing. We get it, you boarded in Ethersworn Cannonist, your opponent was running Kresh and Hellkite Overlord, and now you smash Jund. Glad you shared the new meta-killer, and now we have something to win Worlds with.
All this embarassment can be saved by simply testing properly. My own methods are pretty unorthodox and quite time-consuming, and I'd like to throw in a disclaimer that I'm not saying it's absolutely perfect, but I like to think it's pretty close.
The first step is to simply play your deck. You can play it however you choose, against whoever you want, wherever you want. Given that I can't afford modo and that I dislike boring my friends with my crazy concoctions, I generally start with Magic Workstation.

...
...
...

If any of you are still reading after that last blow to my credibility, yes, I use Workstation, though modo would suffice as well. Quite simply, at this stage, you're just making sure the gears and cogs of your deck actually turn and don't just break and fall off. There's no sense in going further with your testing if your control deck has bad mana, your aggro deck can't beat a wrath effect, or your combo deck goldfishes on turn fourteen. Testing ten or so games on mws, or via any other effort, will make sure you have a foundation for your deck in the first place. I strongly advise playing the usual "two out of three, sideboard" type matches, even this early. And don't tell yourself you need to see how main deck works first, or that you don't know what your weak matches will be. If you can't figure out this elementary theory in a clearly defined format, save yourself the trouble and try a different deck. Right now, in standard, you need to have game against Jund, Vampires, Boros Beats, 3-5c control, and Naya. Build a sideboard accordingly and start testing, you can always adjust it later. Honestly, I can't stress this enough, not playing with a sideboard is about as bad as not playing with lands. At the very worst, let your opponent sideboard after game one even if you can't figure one out, so that you know you're not cold to Pithing Needle, Thought Hemmorage, or other assorted hate cards. Once you've ensured that your deck can function well enough to merit further testing, you're ready to move on.
The second part is a method that you need a lot of patience and mental stamina for. Again, I use mws, but you could conceivably proxy up a gauntlet and do this at home. This is a little controversial, but I find this to absolutely prove beyond any doubt if your deck is good enough: simply load your deck, create a game with a silly title, load as your second deck whatever you feel the best deck in the format (hint: Jund), join your first game, click "accept", and get ready to waste a few hours.
Yes, you're playing Magic against yourself. Yes, this means you see both players hands and make decisions for both, which is entirely contradictory to how Magic is actually played. Yes, I know this sounds awkward. The catch is that you play the Jund deck as if it has absolutely perfect information, including knowledge of what's in your new deck's hand, how it's mana will play out, etc. I personally don't allow the "best deck" to know what's on top of each player's library, because that gets a little confusing, but if you can manage, by all means, do so. Play ten games like this, with Jund playing perfect and your new concoction playing as if it were a normal match. I know your opponent in real life won't always play around Swerve, and I know that allowing them to do so skews results. But hey, if you're still winning with your deck despite this handicap, you're on to something, and how much easier will your real life games be when your opponent does pay three mana and a card to discard two more cards and bolt itself?
Once your ten games are done, take a look at the results. If you're even 50-50 with a handicap this bad, chances are your deck might be doing something right. If you're under 40%, which is where a lot of decks end up, it's not to say that your deck can't beat Jund, especially when the Jund player is susceptible to normal human mistakes and incomplete information, but do really want to play a deck that is 40% against a perfect opponent playing the best deck in the format? What happens to that variance when they're simply playing averagely? Is 50% that much better against a deck you'll probably play against four to five times in a tournament?
If you're happy with the results, sideboard your deck as you normally would with incomplete information, and then sideboard the Jund deck with perfect information, including the knowledge of what your new deck is sideboarding in and out. Play ten more games, using the same method as before, and see how things work out. Pay close attention to additional options you had in sideboarding. For example, post board against Blightning, certain decks are better of drawing than playing. If you missed this and ended up with a poor percentage, feel free to try again with a new approach. Once you're satisfied that you've done all you can or all you're willing to do to fix the matchup, check your results again. Once more, if you're 50% or better, things are looking good.
Take a break for a bit, and when you feel up to it again, do the same thing against the other tier one decks in the format (A sidenote, people seriously need to learn what "tier one" implies. I've quit far too many games because people join with rogue decks -yes I'm a hypocrite- and then claim that they're tier one decks. A rogue deck can smash every deck in the format, it's still not tier one by definition until it becomes an established archtype, and I can assure you, you aren't paving the golden road for it by testing it in random mws games), until you're confident that your deck can hold it's own against a fair amount of the field. If you beat jund, vampires, and blue-based control, but lose to Boros Beats, that's perfectly acceptable. If you smash Jund but lose to everything else, you might have to rethink the value of your deck at a tournament, though it still could be the right call. Afterall, Sanity Grinding scooped to everything that wasn't Tokens but still managed to top 8 events in the right meta, so keep an open mind. Once you're happy with your deck after all this self-testing, you're not only be playing that deck well because of all the testing, but you'll have very good knowledge of how to play the other decks well, and will know where all the holes in their plan are.
Now is the time where you find some skilled friends, some local tournaments, or some modo 8-mans and play your best. I do not recommend mws at this point, as the average player is pretty terrible. Additionally, it helps you have the deck in your actual hands, as going through the motions online is a lot different than real life. Just ask people that tried to play Project X on modo. At this point you'll simply be tuning your deck and your game to be at it's best. You'll probably start to pick up some cool tricks and traps for your opponents that your "perfect" self wouldn't have allowed to happen while testing before. You should be winning the majority of your matches now, and you're simply seeking out all the obscure advantages you can find by testing with other people. If you're doing well, you're ready for anything. At this point, even when people disagree with you, you no longer have to feel like a donk when you start yelling how great of a meta-killer your deck is.
So that's pretty much it. I find there's a lot of value in using my method, and I hope this helps anyone that doubts the accuracy of their results, or has them routinely doubted by others. There's usually a good reason, and hopefully this article helps eliminate it. Thanks for reading.