Off the top I am pleased with the intricate variant design and the high modality of play styles, at least theoretically…
What is clear is cavalry is CORE to gameplay hence the synergy with the Grasslands mechanic which exclusively benefits cavalry. Yet the tuning and functions of the unit roster are far from intuitive (at least not intuitive to my smooth brain 
).
Let’s start with the worst unit first:
Mohe Tribesman
This unit cost more than standard archers and takes longer to train. While the unit is faster than archers and technically kill spearman in one less volley than archers, their kiting is inferior, gimpy, AND spearmen can easily kill this unit especially in extremely high masses where micro is near impossible. Also they trade verses archers 1 to 1 but b/c they are more expensive means archers are a soft cost-effective counter.
This means all 3 base units counter this unit. The singular benefit to this unit is it presents Jin Dynasty a Rus like Knight/Horse Archer comp! Right?! …
Iron Pagoda
If you do not include the cost and training discount of the Grassland, this unit takes too long to train and its splash damage only hits spears archers and villagers as those are the only units that clump up close enough together to even see the splash dmg. Either the training time and cost needs to be reduced or the splash dmg needs a significant buff by extending the radius!! If instead the devs want these units to function like the mongol variant special lancers then the unit needs a lot more HP.
Bed Crossbows
These unit are better than springlads in cost, train time and availability compared to the springald! However, it’s inferior in killing heavy melee ? I still have some ideas to try out with this unit, so I think it’s still promising.
Before I get on to the next point I do like most all the Jin units can’t be single massed in the same french knights or desert raider can be at times.
Economy
This is technically by far the most lackluster aspect about this civ in terms of sheer momentum. Building mounted villagers on paper eclipses building standard villagers in about 7mins of gameplay! However no civ essentially has “standard” villagers, every civ has an eco gimmick; for example the tughlaq villager which has a modest 5% baked in gathering bonus and on paper the tughlaq eco cache will always outpace the mounted villagers resource???
It gets worse. Bc these villagers cost gold and they consume resources twice as fast, it means this civ gets pushed out to unsafe resources much faster than any other civ, while not netting an eco advantage for sometime if ever??? Imagine that 2 civs are effectively in lock step in resources cache but the one civ has to transition to 2nd deer pack in half the time (with MEH 
units at that?)???
If they left all the units above as is but made this eco POP, out the gate?? Idk… right now a mounted villager is worth in most cases 2.2 standard villagers but is trained every 35s. Compare that and their cost to standard villagers and you have a situation of the mounted villagers are 25.7% more efficient at gathering per second it takes to train a mounted villager compared to standard villagers. However… Mounted villagers COST 35% more resources per second it takes to train compared to a standard villagers. So just on paper the mounted villagers has to over come that before they can really excel over the standard villager. THEN REMEMBER TO FACTOR IN THAT no civ has JUST standard villagers as their complete eco…means mounted villagers have an even bigger hill to climb and they need to climb that hill by a cap of 20 (26)???
Here is a Gemini generated graph of 20min of eco from a few different villager types. Keep in mind I made the food infinite under the TC and gold infinite at 9 tiles away. Things that are clearly untrue…yet check this out.
OTD villagers cache takes the lead followed by tughlaq then mounted. Mind you this doesn’t factor in wheel barrow nor eco upgrades and the Jin dynasty villager STOP training mounted villagers at 20 (26 in this graph) before reverting to standard villager production hence why you can see it tapering off. Regardless, the therortical, yet highly unrealistic eco, still puts jin cache eco at 3rd, which in reality is worse given eco bonuses not considered (upgrades and other niche civ benefits) and resources not being infinite near the TC…
I’ll edit the post once my limits are reset , but I wanna have a mounted villager eco cached that exceeds all others by minute 8 and that be upfront benefit of this civ, that others need to slow you down and or pick you apart once you go out onto resources. And I want that achieved by either a cost reduction or gathering rate increase if not both.
UPDATE:
new graph with the same unrealistic stipulations above however I asked gemini to reduce the cost and buff the gather rate slightly to allow mounted villagers to take the eco cache lead by 8mins yet not dwarf the 2nd place eco (OTD) by more than 10% by the 20min mark shown.
it reduced the cost by 5 food and 5 gold and then boost the raw gathering rate by ~18%. If sure the dev have better simulators where they can better dial in the effectiveness of this villager type.