Right, that doesn't look like too terrible a task
. Outline plan is:
1. Obtain mame driver for platform
https://github.com/mamedev/mame/blob/master/src/mame/drivers/taito_f2.cpp
2. Have a look to see whether Hat Trick Hero (aka Euro Football Champ aka EFC) has any weird hardware.
I/O Priority / Palette Additional gfx Other
footchmp TE7750 TC0360PRI TC0260DAR TC0480SCP(tilemaps) TC0190FMC(4 players input?sprite banking?)
3. Decide that doesn't look too weird, other than being a minority on the platform by using TE7750 for IO. I don't know what that does at this point so I'm not going to worry about it and instead assume it'll be fixable later.
4. Have a look to see what desirable thing runs on F2 and has similar hardware. Liquid Kids?
liquidk TC0220IOC TC0360PRI TC0260DAR
5. So this looks feasible: palette and priority silicon is the same and it doesn't need the extra tilemap IC so we can probably find some values to poke its registers with to get it to shut up. Or perhaps not and then we could get magic trees appearing like some RType'd Major Title conversions apparently did.
6. Look at the address space maps for both games to see whether they have similar amounts of RAM/ROM, whether it's in similar locations and where stuff like IO lives and custom chip registers. Whilst scrolling down notice some info on sprite triggering & bank switching and see there is some similarity (irq6 is used as a simple flag, something else does the magic) so not getting too worried about this yet. The address space layout will be the biggest indicator as to whether this is feasible. It might match exactly, we might have to cut a few traces and patch some pins to rearrange the address space, or we could find nothing overlaps and we have to patch the ROM/PCB loads to get anywhere.
[font="Courier New, Courier, mono"]static ADDRESS_MAP_START( footchmp_map, AS_PROGRAM, 16, taitof2_state )[/font]
[font="Courier New, Courier, mono"]
AM_RANGE(0x000000, 0x07ffff) AM_ROM[/font]
AM_RANGE(0x100000, 0x10ffff) AM_RAM
[font="Courier New, Courier, mono"]
AM_RANGE(0x200000, 0x20ffff) AM_RAM AM_SHARE("spriteram")[/font]
[font="Courier New, Courier, mono"] AM_RANGE(0x300000, 0x30000f) AM_WRITE(taitof2_spritebank_w) /* updated at $a6e, off irq5 */[/font]
[font="Courier New, Courier, mono"] AM_RANGE(0x400000, 0x40ffff) AM_DEVREADWRITE("tc0480scp", tc0480scp_device, word_r, word_w) /* tilemaps */[/font]
[font="Courier New, Courier, mono"] AM_RANGE(0x430000, 0x43002f) AM_DEVREADWRITE("tc0480scp", tc0480scp_device, ctrl_word_r, ctrl_word_w)[/font]
[font="Courier New, Courier, mono"] AM_RANGE(0x500000, 0x50001f) AM_DEVWRITE8("tc0360pri", tc0360pri_device, write, 0x00ff) /* 500002 written like a watchdog?! */[/font]
[font="Courier New, Courier, mono"]
AM_RANGE(0x600000, 0x601fff) AM_RAM_DEVWRITE("palette", palette_device, write) AM_SHARE("palette")[/font]
[font="Courier New, Courier, mono"] AM_RANGE(0x700006, 0x700007) AM_WRITE(taitof2_4p_coin_word_w)[/font]
[font="Courier New, Courier, mono"] AM_RANGE(0x700000, 0x700001) AM_READ_PORT("DSWA")[/font]
[font="Courier New, Courier, mono"] AM_RANGE(0x700002, 0x700003) AM_READ_PORT("DSWB")[/font]
[font="Courier New, Courier, mono"] AM_RANGE(0x700004, 0x700005) AM_READ_PORT("IN2")[/font]
[font="Courier New, Courier, mono"] AM_RANGE(0x70000a, 0x70000b) AM_READ_PORT("IN0")[/font]
[font="Courier New, Courier, mono"] AM_RANGE(0x70000c, 0x70000d) AM_READ_PORT("IN1")[/font]
[font="Courier New, Courier, mono"] AM_RANGE(0x70000e, 0x70000f) AM_READ_PORT("IN3")[/font]
[font="Courier New, Courier, mono"] AM_RANGE(0x700010, 0x700011) AM_READ_PORT("IN4")[/font]
[font="Courier New, Courier, mono"] AM_RANGE(0x800000, 0x800001) AM_DEVWRITE("watchdog", watchdog_timer_device, reset16_w) /* ??? */[/font]
[font="Courier New, Courier, mono"]
AM_RANGE(0xa00000, 0xa00001) AM_DEVWRITE8("tc0140syt", tc0140syt_device, master_port_w, 0x00ff)[/font]
AM_RANGE(0xa00002, 0xa00003) AM_DEVREADWRITE8("tc0140syt", tc0140syt_device, master_comm_r, master_comm_w, 0x00ff)
[font="Courier New, Courier, mono"]ADDRESS_MAP_END[/font]
And the map for Liquid Kids:
[font="Courier New, Courier, mono"]static ADDRESS_MAP_START( liquidk_map, AS_PROGRAM, 16, taitof2_state )[/font]
[font="Courier New, Courier, mono"] AM_RANGE(0x000000, 0x07ffff) AM_ROM[/font]
[font="Courier New, Courier, mono"] AM_RANGE(0x100000, 0x10ffff) AM_RAM[/font]
[font="Courier New, Courier, mono"] AM_RANGE(0x200000, 0x201fff) AM_RAM_DEVWRITE("palette", palette_device, write) AM_SHARE("palette")[/font]
[font="Courier New, Courier, mono"] AM_RANGE(0x300000, 0x30000f) AM_DEVREADWRITE8("tc0220ioc", tc0220ioc_device, read, write, 0x00ff)[/font]
[font="Courier New, Courier, mono"] AM_RANGE(0x320000, 0x320001) AM_DEVWRITE8("tc0140syt", tc0140syt_device, master_port_w, 0x00ff)[/font]
[font="Courier New, Courier, mono"] AM_RANGE(0x320002, 0x320003) AM_DEVREADWRITE8("tc0140syt", tc0140syt_device, master_comm_r, master_comm_w, 0x00ff)[/font]
[font="Courier New, Courier, mono"] AM_RANGE(0x800000, 0x80ffff) AM_DEVREADWRITE("tc0100scn", tc0100scn_device, word_r, word_w) /* tilemaps */[/font]
[font="Courier New, Courier, mono"] AM_RANGE(0x820000, 0x82000f) AM_DEVREADWRITE("tc0100scn", tc0100scn_device, ctrl_word_r, ctrl_word_w)[/font]
[font="Courier New, Courier, mono"] AM_RANGE(0x900000, 0x90ffff) AM_RAM AM_SHARE("spriteram")[/font]
[font="Courier New, Courier, mono"] AM_RANGE(0xb00000, 0xb0001f) AM_DEVWRITE8("tc0360pri", tc0360pri_device, write, 0x00ff) /* ?? */[/font]
[font="Courier New, Courier, mono"]ADDRESS_MAP_END[/font]
[font="Courier New, Courier, mono"]
[/font]
[font="Courier New, Courier, mono"]
[/font]
[font="Courier New, Courier, mono"]
[/font]
Aaaand we find we have some 'challenges'. Green is common between the pair and should be fine, purple is common hardware but different addresses - can we cut & patch traces or do we look at adding logic to fiddle the addresses, can we even get to this stuff or is any of it integrated inside customs?
[font="Courier New, Courier, mono"]
[/font]
It's about now we wonder whether converting it is really a good idea

We have the RAM and ROM in the same space (but the ROM was almost inevitable due to how a 68k boots and gets its vectors) and we have some direct matching hardware that is mapped at a different range. We also have a mismatch on the tilemap IC but maybe the mame drivers will reassure us the minimal features of the tc0100scn used by Liquid Kids exist in the fancier tc0480scp of EFC. But the control registers are much larger on the 480 so maybe we're in for disappointment!
I can't find any F2 boards here in a 5 minute search so without seeing the PCB I'm assuming we're going to try to fix this in software. So for the matching but mis-mapped hardware we'd need to patch all address references in the liquid kids ROM to use the new offsets. Some of this will probably be quite straightforward. IDAPro does a pretty good auto initial disassembly and if we're lucky there might be some obvious references to the base of sprite ram and the sprite custom. But there is no automated way of scanning through the ROM and determining which occurrences of 0x90xxxx are references to sprite RAM, data or sequences of machine code so we might find some sprite handling routines but miss others.
So we'd either have to patch a version of mame to track the PC when any write to spriteram happens, go to the disassembly and identify the routine and patch it - and then hope our playthrough caught all update methods - or disassemble the entire ROM to see where any references live and patch there.
Then we have the other hardware to deal with

palette, tilemap control etc. At this point I've not even looked to see if the graphics data structure is identical between the boards so we may or may not be lucky there.
By now I'm already thinking Liquid Kids is an ok game, but not good enough that I'm willing to go to this effort

Perhaps a better approach would have been to compare address space first to narrow down the candidates, but Liquid Kids was the only game on the platform I liked the look of.