I think there is a point being missed if you want to increase the difficulty of 5e. It would be better, IMO, to address the basic math of the game to increase difficulty, rather than trying to do it on the back end with increasing encounter difficulty or bolting on new rules.
I'll start with direct responses and then give some more general thoughts.
1. Reduce Proficiency Bonus for PC's by 2. So, a 1st level PC has a Proficiency bonus of 0 and a 20th level PC has a +5. This has all sorts of knock on effects - caster save DC's become a lot lower, attacks miss more often and PC's fail saving throws more often. Drop the bonus by 1 instead of 2 for a lesser effect.
I've found that if you do 2., there is no need for 1.
2. Reduce the Standard Array. Instead of 15,14,13,12,10,8, change it to 13 12 12 10 10 8. With racial bumps, that means the highest stat a PC will have is 15. Combine that with the reduced Proficiency bonus and suddenly the PC's are MUCH weaker. We're talking a DC of 10for the best caster and a +2 attack bonus at 1st level. And that reduction will affect play all the way through. And, let's not forget, that takes 3 ASI's to reach 20 - 12th level for a lot of characters.
Another way to achieve this is to use points-buy but remove the stat bumps for species. One can also reduce the points spent, as you effectively do. In your example, you reduce it to 17! But one need not go quite that far.
3. Drop Hit Dice by one step. D10 becomes D8, d8 becomes d6. Not a huge drop, but, again, combined with a lower Standard Array, and suddenly your PC will only have 20 HP at 5th level.
I like this. Another approach is to stop giving HD with level progression, but that over-penalizes classes whose effective-feats (you can value classes in half-feats, as I (and others) have shown elsewhere) are invested in hit dice. Your approach seems better at first glance.
Those three simple changes will drastically raise the difficulty of the game.
Player options will become pinched when they fail rolls more often and have fewer hit points (equating to lessened tempo). That does increase experienced difficulty, and it forces players to make correct choices (or die). Let's call this the accuracy and tempo axis of difficulty.
Another axis of difficulty is the complexity of decisions. A big meat-sack with a hard-hitting attack has low decision-complexity even if it is going to almost certainly defeat you. You'll die, but not in a very interesting way. Changing that to six foes, half of whom have say crossbows with Crossbow Expert feat, two of whom have Defense fighting style, splint and shield, and one of which is a caster, has far higher decision-complexity. It seems better to take out the caster first, but those crossbows will soon hurt. We should ignore the tanks, but what if they have a reasonable attack like polearms with Polearm Master? What if the caster is also a tank?
Base 5e is low difficulty in at least these two ways. Character hit points and success rates are good enough that players can make inaccurate decisions and still succeed. They enjoy an effective buffer against their mistakes. Many monsters are overly simplistic, and the encounter XP thresholds procedure over-costs larger groups of foes. Many of the most interesting and complex-decision forcing mechanics are reserved for player characters... making it fairly obvious that breaking that privilege offers opportunities to challenge players more.