The 50 substrings that validate any string of Roman numerals [FINALIZED]
Now posted: The 50 substrings that validate any string of Roman numerals
Given a string of Roman numerals, decide whether it forms a valid Roman number. If not, output the substring that proves this, from the list of 50 strings described below.
Relevant fact
This challenge is based around the following fact:
A string of Roman numerals is a valid Roman number if and only if it contains none of the following 50 strings as a substring:
"CCCC", "CCD", "CCM", "CDC", "CMC", "CMD", "CMM", "DCD", "DCM", "DD", "DM", "IC", "ID", "IIII", "IIV", "IIX", "IL", "IM", "IVI", "IXC", "IXI", "IXL", "IXV", "IXX", "LC", "LD", "LL", "LM", "LXC", "LXL", "MMMM", "VC", "VD", "VIV", "VIX", "VL", "VM", "VV", "VX", "XCC", "XCD", "XCL", "XCM", "XCX", "XD", "XLX", "XM", "XXC", "XXL", "XXXX"
This works for any length string, if a valid Roman number is defined as follows:
Valid Roman numbers^{[1]}
 Each numeral appears no more than 3 times consecutively.
 Each of
V
(5),L
(50), andD
(500) appears no more than once consecutively.  A Roman number is constructed by concatenating the strings representing its thousands, hundreds, tens, and units components.
Decimal  Thousands  Hundreds  Tens  Units 

1  M  C  X  I 
2  MM  CC  XX  II 
3  MMM  CCC  XXX  III 
4  CD  XL  IV  
5  D  L  V  
6  DC  LX  VI  
7  DCC  LXX  VII  
8  DCCC  LXXX  VIII  
9  CM  XC  IX 
So, for example, 2345 would be represented as the concatenation of MM
(for 2000), CCC
(for 300), XL
(for 40), and V
(for 5), or MMCCCXLV
.
This defines a unique correct representation for each number from 1 to 3999 (4000 and above are not representable without breaking the first two rules). The 50 substrings method described above will identify each of these 3999 strings as valid, and all other strings of Roman numerals as invalid.
Input
 A string containing only Roman numerals (
I
,V
,X
,L
,C
,D
,M
).  The string will have length at least 1.
 The string will have length at most 15 (this is the length of the longest valid string of Roman numerals).
Output
 If the input is a valid Roman number, output a consistent value indicating this.
 Consistent means that the value must be the same for all valid Roman numbers.
 The output for a valid Roman number must not be one of the strings from the list of 50.
 If the input is not a valid Roman number, output exactly 1 string from the list of 50.
 The output in this case must be a substring of the input.
 If the input has 2 or more of the strings from the list of 50 as substrings, you may choose any 1 of them to be the output, but you must choose only 1 of them (you must not output 2 or more).
Examples
A valid Roman number
The input MCMXCVI
is the unique correct representation of 1996. It contains none of the 50 strings.
A string that is not a valid Roman number
Although the input MMXDIII
might be suspected of representing 2493, it is not the unique correct representation of this number (which is MMCDXCIII
). Note that it has XD
as a substring, identifying it as invalid. The only correct output is therefore XD
.
An invalid string with more than 1 potential output
The input MMCCMDXXV
has 2 substrings that make it invalid, so either CCM
or CMD
would be correct outputs. It would not be correct to output both of these, or to output their overlap CCMD
, as this is not one of the 50 strings.
Test cases
Test cases are in the format INPUT : VALID, OUTPUTS
. Note that only one of the valid outputs can be chosen  outputting 2 or more is incorrect.
I : VALID
V : VALID
X : VALID
L : VALID
C : VALID
D : VALID
M : VALID
II : VALID
VV : VV
XX : VALID
LL : LL
CC : VALID
DD : DD
MM : VALID
III : VALID
VII : VALID
IVI : IVI
IIV : IIV
CCI : VALID
CCV : VALID
CCX : VALID
CCL : VALID
CCC : VALID
CCD : CCD
CCM : CCM
IIII : IIII
MLDI : LD
MXXC : XXC
DCIIX : IIX
MCXXXX : XXXX
MCCCCXVI : CCCC
MMLXCVII : LXC
MMMCMXCIX : VALID
MMMDCCCLXXXVIII : VALID
MMCCCXLV : VALID
MCMXCVI : VALID
MMXDIII : XD
MMCDXCIII : VALID
MMCCMDXXV : CCM, CMD
XXX : VALID
CLLX : LL
DXXDMMV : DM, XD
CCDDDIMDD : DD, IM, CCD
VLCXIVXMCVXLC : VX, VL, LC, XM
DVLIILVCXVXVMLI : VX, VL, IL, VM, VC
VVDLMIVILXXDX : VV, VD, IL, XD, LM, IVI
DMXXCMILVCMLLMV : DM, IL, LL, VC, LM, XXC, XCM
CMXDVLCCDDLXLXC : DD, VL, LC, XD, XLX, LXC, CCD, LXL
XDIXCLLMVVLCMCM : VV, VL, LC, XD, LL, LM, IXC, CMC, XCL
IIXXCDVVLMILVDD : DD, VV, VL, VD, IL, LM, XXC, IXX, IIX, XCD
DDMIIXXCMCCDCMM : DD, DM, CMC, CMM, XXC, IXX, DCM, CDC, XCM, CCD, IIX
Scoring
This is a code golf challenge. Your score is the number of bytes in your code. Lowest score for each language wins.
Explanations are optional, but I'm more likely to upvote answers that have one.

This is a common modern set of rules, described as Standard form on Wikipedia. It does not reflect all usages during history, but will be the basis of this challenge, since otherwise the 50 substrings approach does not work. ↩︎
1 comment thread